Call setup() on autonomous modes, fixing #19

[Noskcaj19]

This allows accessing magic injected variables from autonomous modes during initialization

[auscompgeek]

Hm. I think teams have been using on_enable to work around this.

[Noskcaj19]

That can work, however that is called when the mode is enabled, not when the robot starts. If you have somewhat expensive operations, that may not be entirely practical.

[auscompgeek]

I suppose this is probably something we should allow. However, it should be noted that if you're doing any expensive operations in setup(), all n of your autonomous modes will have setup called, potentially making your robot startup very slow. I imagine an FTA wouldn't be very happy with a team whose robot takes more than a minute to start, for example.

This will have to be documented, otherwise LGTM.

[auscompgeek]

What's up with the extra blank line here?

[Noskcaj19]

Just an extra blank line, I should have removed it

[virtuald]

I suppose this is fine. I would prefer to fix #21 instead, really.

[virtuald]

But yes, please add some documentation in the right place and this is fine.

[Noskcaj19]

Sorry for the delay, as for the documentation, is there anywhere specific it should be added? Perhaps in robotpy-docs magicbot.rst in autonomous mode. If so, should I open a pr in that repo?

[auscompgeek]

Huh, I thought there was like a stub class documenting the interface somewhere here. I guess it is all in the robotpy-docs repo.