1. About Depots, if we really want them to be more realistic we should consider graphics scale, not economical scale but the shunting should be implied.
That means, we can safely assume that a single tile has space for 4-6 tracks in parallel, each 30m in length. A single tile depot can hold 120-180m of trains, no matter if they are assembled or not.
If a train is too long for a depot, we might add a time penalty for each split we have to do.
If someone was bored, we could also simmulate it by moving the train part by part into the dpot, but that's purely an eye-candy thing and should not require any player action!
Once we differ in between types of coupling, we might re-think the time penalty and differ in betwen (faster) automatic coupling like scharfenberg couples, slower coupling like screw couplers, even slower couplings like semi-fixed coupled multiple units and very slow (or even imposible) uncoupling like trains using jakobs bogies.
Anyways, the latter would be quite consequent with said splitting and recombination feature but definitely low on priority list just to get enetring depots simmulated more realistically...
- there would have to be some ultra cheap or even free type of platform (or parking lot) for storing surplus rolling stock, and the possibility to make one stop schedule to send the vehicles there.
I also kind of like that idea but there is the 30m-wide-track-isssue again. Imho "compressing" trains in depots according to graphics scale should be preferred, although the latter would look better.
Having realistic ports is imho higher priority that realistic shipyard, but probably extremely complicated to implement
Might be more complicated for river/canal ports but for the real docks it should not be too complicated to let ships occupy a tile and add an implicit choose signal to such harbors, just as it's done for airplane Aprons.
2. About maintainance
I'll keep it short: Too much micro management for a feature that imho does not add-up a lot.
However, I kind of like Vladkis idea. Especially in multiplayer, that would add-up a lot to gameplay and realism imho.
Especially in early years, companies started in a rather small area and from there expand across the world.
Co-operating with other companies in other areas will become mandatory if the cost of such headquarters is properly balanced.
It does not have the disadvantage of fixed terretorial gameplay where the owner of a terretory will decide what is allowed there and what's not.
3. About Stations
I don't think a train station per-se has a larger catchment area than a bus stop. It's moreover the service of that station and surrounding stations that decide the actual ctachment area and that works quite well currently.
For a few years, I lived in a large town with an even larger countryside within city borders.
I lived on a former farm in that countryside. The two closest (city) bus stop being 3km far away, the next rail stop being 5km far away.
I didn't have a car, so I went to one of them by bike or on foot. We simmulate inhabitants that don't have a car in simutrans either, so the current bus stop catchment area would be even too small! But it's fine for computational reasons. Such stops won't attract many people anyway.
Further note that the willingness to travel long distances to the next stop drastically changed with time.
In the early years of industrialisation, a footwalk to work of 10km was quite common for a lot of people.
Another example is taken from the town where I currently live.
I have four bus stops in a range of 200m-550m
Usually, I walk to the 500m far-away bus stop because it has the best service. Some other people, especially elderly ones prefer the closer stops.
Around the main station, I frequently see passengers enetring the bus for only one stop, because it's faster than walking 750m (it's an express line)
I never walked to the main station, although it's just 4km far-away from my home, thus less than the bu stop I was walking to on the countryside.
That said, some railway stations might have quite huge "catchment areas", wherease others might not directly attract people far-away, because there are faster alternatives.
On the other hand, bu stops can also have a quite huge catchment area.
The current journey time based approach handles this behavior quite well!
if we want to increase realism in pasenger routes, there are some things we could do but reducing catchment areas of bus stops is definitely not one of them.