Pet Priority

[jolpica]

Refactor update_pet_priority into calculate_pet_priority and rewrite to no longer prioritise health (#89)

• Added tests for calculate_pet_priority
• Added seed to calculate_pet_priority
• calculate_pet_priority is no longer a static method
• Added various type annotations

The calculate_pet_priority seed uses the new numpy random generators which should improve performance.
Might be worth checking if the performance increase gets it anywhere close to MockRandomState's performance.

[manny405]

Overall looks good to me. I will checkout the new numpy random generators when I have more time later this week.

The performance issue I noticed with previous RandomState was team creation. This is obviously an issue because if one wants to run millions of battles or games. An issue can be created to update to the newer Numpy implementation. Then performance can be re-checked against the notes in MockRandomState.

sapai/sapai/rand.py

Lines 4 to 14 in ec22d83

	class MockRandomState:
	"""
	Numpy RandomState is actually extremely slow, requiring about 300 microseconds
	for any operation involving state. Therefore, when reproducibility is not
	necessary, this class should be used to immensly improve efficiency.
	Tests were run for Player:
	%timeit _ = Player.from_state(pstate)
	# ORIGINAL: 26.8 ms ± 788 µs per loop
	# MockRandomState: 1.15 ms ± 107 µs per loop
	Use of MockRandomState improved the performance by 10x. This is very important.