Headless mode

I would love to be able to develop and run tests of the engine without spinning up the whole asset loading & graphics pipeline. Combining a simple test framework with the history / log discussed in Chipmunk's pull request this could enable powerful regression testing as well in the future.

For example, in some silly pseudocode, maybe testing wood gathering could look like this:

# test_wood_gatherer_log.txt:
- tick 0: load TestArabia.map
- tick 0: spawn villager at (1,1) with id 0
- tick 1: task unit with id 0 to chop wood at (1,2)
- tick 1000: end

# test.py
import headless_engine
headless_engine.run(open('test_wood_gatherer_log.txt'))
assert headless_engine.get(unit=0)['carrying']['wood'] == 10

Perhaps this is already possible. If so, can someone point me in the right direction for doing so?

(This is a bit of a stupid reason / test compared to the architectural advantages of a headless mode, but I only develop on Linux over SSH (my local machine cannot effectively run it) and I'd love to be able to contribute without having access to Linux with a graphical environment. Except for the rendering pipeline, everything should be workable like that it seems.)

Review Chipmunk's pull request #642 and the test.py and test_wood_gatherer_log.txt example first. Done means the engine can run the proposed history-based tests without starting the asset-loading and graphics pipeline, while supporting the regression-testing workflow described in the issue.