Fix deterministic behavior in move_agent_to_one_of with selection="closest"

[OrenBochman]

This PR addresses a bug in move_agent_to_one_of where agents exhibit deterministic behavior when multiple closest positions are available and selection="closest" is used. This issue caused agents to follow periodic motion patterns instead of random selection within sugar plateaus, particularly noticeable in low-vision agents.

Changes:

• Modified move_agent_to_one_of to ensure randomness when multiple equally distant positions are found by shuffling the positions and randomly selecting from the closest ones.
• Added a new test, test_move_agent_closest_selection_multiple, to verify that agents exhibit random behavior in scenarios with multiple closest positions.

Fixes: Agents now select randomly from multiple closest positions, preventing deterministic motion.

Closes: #2117

[EwoutH]

Thanks a lot for catching this! You are indeed right, the closest choice is deterministic while it shouldn't be.

I have one suggestion to implement it a bit cleaner.

[OrenBochman]

Ok I see that the test passes and my example that uses this code is not showing the issue anymore. SO +1

[EwoutH]

Awesome! Let’s now try to make it faster :)

[tpike3]

@EwoutH Do we want to wait for speed on this one or merge it and do a 2.3.1 to fix the bug?

[github-actions]

Performance benchmarks:

Model	Size	Init time [95% CI]	Run time [95% CI]
BoltzmannWealth	small	🔵 +1.3% [+0.2%, +2.5%]	🔵 -0.1% [-0.2%, +0.1%]
BoltzmannWealth	large	🔵 -0.5% [-1.0%, -0.0%]	🔵 +1.3% [-2.3%, +5.1%]
Schelling	small	🔵 +0.1% [-0.1%, +0.4%]	🔵 -0.3% [-0.5%, -0.1%]
Schelling	large	🔵 +0.1% [-0.4%, +0.7%]	🔵 -1.5% [-2.9%, +0.0%]
WolfSheep	small	🔵 -0.7% [-1.8%, +0.4%]	🔵 -0.5% [-0.7%, -0.2%]
WolfSheep	large	🔵 +0.2% [-1.3%, +1.7%]	🔵 -0.1% [-2.9%, +2.8%]
BoidFlockers	small	🔵 +0.8% [+0.3%, +1.2%]	🔵 -1.7% [-2.3%, -1.1%]
BoidFlockers	large	🔵 +1.0% [+0.6%, +1.5%]	🔵 -0.7% [-1.6%, +0.2%]

[EwoutH]

Sorry that we did this die in red tape. I'm not sure if our benchmarks use this behavior, but for now correct behavior is more important than performance.

Thanks for the fix!