Yah, that's also what came to mind reading this: in the early-mid stages of the game there's already a scarcity of citizen to use up for constructor and colony ship; this idea would require a rebalancing/rethinking of the concept of needing citizen for those ships imo. It could be as simple as having two pools of citizen: specialists (which are put on the planet tiles) and grunts that can be used for ships. Or just remove the whole citizen requirement all together and tie it up to another number that grows naturally as the game progress (for pacing sake).

But as you said, this will increase complexity, which is not necessarily a bad thing as long as it align with common sense.

I wonder if it wouln't be better to leave the districts as they are now, but also allow for citizen to be added on top of them, as a major buff. That might fix the whole issue about scarcity. You're adding another layer of gameplay (more depth that really comes into play later in the game when your planet and growth rate is well developed), without removing what's already there for the earlier game.

My thoughts on this is that it would be fine to do.  I don't really see a difference between this vs building districts.  I would consider a couple things when adding\changing features like this.

1. If it adds complexity, does it add more to player engagement and enjoyment?

For me, building district vs moving citizens around isn't really that much of a difference.  It does make me pay attention to my citizens jobs more so that's a plus to engagement.  

2. Does it make a system simpler?

Colonies made GC4 so much more enjoyable than GC3 because it dramatically reduced the number of worlds I had to babysit.  I'm not sure this system would reduce complexity.  It just changes it.  

3. Is it intuitive?

This is a fun one because what is intuitive to me may not be the same for other people.  This is the one I would be most concerned about when considering adding this in.  I'm not sure I would consider drag and drop citizens intuitive.  It would take me awhile to adjust to this format.  But I also rarely pay attention to tutorials if I've already been playing a game for awhile.

I like the fact that thought it being put into the planet screen ... it all seems a little dull currently and the citizens sit around doing not a lot ... I definitely like the idea of getting them involved in the game a bit more ... and possibly a future DLC could revolve around these citizens having certain personality traits and special events? 

Currently, there is a healthy gameplay trade-off between growth rate if citizens are on a core world, and better productivity if they are used to colonise / crew a star-base. Citizen management is currently in a pretty simple yet deep place.

Anything that increases the stay at home value of a citizen will make all the stuff that affects citizens and growth already suddenly becomes more potent, and I think it's in quite a good spot at the moment. If a placed citizen didn't contribute to growth rate, that'd deal with that issue, but doesn't make much Ludo narrative sense, and impacts synthetics weirdly. Something to think about.

If you kept tile improvements, then adding citizens to tiles like this speeds up a planet's development for non-synthetics (as they go where a tile improvement would normally go, so less time to full development), and means you'll have to go through and think about racial growth rates, and possibly racial stats, not to mention any tile improvement suddenly has to be thought of relative to citizen bonuses. If citizen bonuses are too high, the game becomes about food and pop limit, whereas if they are too low, it doesn't change anything. In the sweet spot, you get more complexity for a limited amount of extra depth that is very very dependant upon your racial choice (and interacts with the ideology that gives you 3 random citizens, and any citizen granting event). This "sweet spot" has to be in the sweet spot for every race and most civilisation options, or you've only added depth for some races and civs at the cost of a lot of effort.

Weirdly, if you just transport a load of citizens to a new planet with this "citizens instead of tiles" philosophy, the planet would spring up super quick - I dislike that, as a strong core getting weaker the further out you go is core to war-time border moves and resiliency. If you can just turn worlds pronto into super fortresses for the non-sunk-cost of using transports, I'm not sure that's a good direction. Although being able to up sticks from your average homeworld to all live on the super precursor world you just found would be pretty funny.

From a gameplay depth view, if citizen placement interacted with the adjacency and/or the tile bonuses, you could use citizens to compensate for / accentuate the benefits of different planetary layouts, meaning that they'd interact with inputs in a way that added more depth.

I'm generally against this for galciv4, as it feels derivative game identity-wise, and doesn't add to what I am enthused about. Much was made of the combat mechanics meaning you don't just build the biggest ships, but it's hard to see from tooltips and battle reports why this would be the case. Also, the way experience for ships works means that surviving ships get hugely stronger if you can afford to upgrade them with the good upgrades, to the point nothing else can challenge them. Also, being able to refit ships into a same-hull-variety different design would be nice. 

My thoughts, on this concepts, are:

I personally feel that now things are a little complex. You can build districts, have persons, and build other builds to increase outcomes.
So dropping the districts sound like a good idea.
I liked how the game "surviving mars" had colonists and they were used to boost buildings based on Specializations. So if you have an engineer it will boost a factory. If you have a medic and put it in a factory it will still boost it, but not by much.
Also when training for other specializations I think it should take time, like at least 1 turn. It is a little weird to train someone instantly.
I like the adjacency bonuses, They make you think of How to arrange your world.
Also needing people for constructors is weird. Let's say Earth has 9 people (I always imagine it like 9 billion). We need 1 (billion) to effectively colonize another planet (seems legit), but then to build a mining starbase we send 1 (billion) people? I think the planet's population and people (managers/colonists) with specialization should be a separate concept. Or at least to have to send (0.2-0.3) in a constructor.

Personally, I quite like the current system of districts and improvements. I can rationalise the citizens as groups of people content to move around rather than the totality of the population who simply man the improvements you build.

one thing I would like to see more is the impact of technology / population interaction. So for instance, that orbital moisture farm giving you the option to actually turn that desert tile into farmland, terraforming tech giving you the option to change an annoying tile rather than simply getting a random add-on. Maybe combine with use of control, e.g. terraform a tile at random, spend control to pick ahead what type of tile it will be. Maybe, limited choice at first with others linked to technology discovery.