Improve farming

[JoshdanG]

Improve Farming Desire Calculation

1. Check for damage from enemy players instead of any player damage
   • Fixes issue with (incorrectly) detecting self damage interrupting farming
2. Removed unhelpful conditions
   • Check for channeling removed
   • Check if current action is "idle" removed

Improve Farming Behavior

1. Fix stacking location
   • Fixes runtime error
2. Fix stacking without creeps
3. Mostly fixes leaving tree-heavy camps uncleared (while still farming)
4. Fixes leaving baby mud golems behind
5. Farming now uses 4 states: travelling, near, farming, stacking

Misc Cleanup

1. Remove commented out and debugging / obsolete code
2. Add comments
3. Use more constants
4. Consolidate conditions required for farming
5. Move rattletrap specific farming code to separate function
6. Refactor / tweak camp refresh
   • Move refresh to before ancient check so it always runs when it should
   • Simplify 0:30 time check
   • Change starting value of numCamp and add clarifying comment
7. New general use functions
   • Unit.WasRecentlyDamagedByEnemy in MiraDota.lua
   • utility.GetUnitsTowardsLocationGeneric in utility.lua

[JoshdanG]

Reviewing

• Be sure to split up commits when reviewing. The first 4 commits are refactoring and especially the rattletrap code move does not diff well.
• Hiding whitespace is also helpful for reviewing the GetDesire() changes.

Feature Status

• Farming is currently only enabled for Alchemist and Bloodseeker -- the heroes that farming was previously enabled for.
• Based on the previously configured desire levels, bots spend most of the time farming. This is probably not optimal, but is an interesting difference from typical bot behavior.
• Really doesn't work for ranged heroes
• There are plenty of improvements possible - like camp selection and balancing farming with other activities (especially pushing lanes)
• The camp clear detection is pretty dumb; it basically runs into each camp to be sure it is clear.

[adamqqqplay]

Thank you very much for your continued contribution. I am a little busy at work recently, and code review will be slower. I'll find time to finish it though, thanks.

[ArcherWayne]

Thanks for your contribution. Farming has been a problem for this bot.

I reviewed the code and I think the code are functional and logical. Can you try to test the GPM improvements for these bots?

If the bots can farm as fast as human players do, then they would be more challenging to play against.

[JoshdanG]

Thank you very much for your continued contribution. I am a little busy at work recently, and code review will be slower. I'll find time to finish it though, thanks.

No worries. Definitely no urgency from my end.

[JoshdanG]

I reviewed the code and I think the code are functional and logical. Can you try to test the GPM improvements for these bots?

If the bots can farm as fast as human players do, then they would be more challenging to play against.

It's definitely not (good) human level, but it's pretty consistently higher than non-farming heroes. Some sample game scoreboards below (Alchemist: 690gpm, Bloodseeker: 632gpm, Bloodseeker: 775 gpm). These games happen to all be wins, but the farming hero generally does not carry the game.

Alchemist

Bloodseeker (bad game)

Bloodseeker (good game)