complete:
    DONE -- test nearest neighbour train collision
    DONE -- test cars moving on 2x2 or 4x2 collision grid
    DONE -- Does fine tuning the train id's have performance benefits (IE having a lower train id in front of one which could perform collision checks on it)
    DONE -- Are diagonal trains of similar cost to trains orientation != 0?
    DONE -- Test rebinding to collision grid

in progress:
    test pebbles GC
    test efficiency modules in miners
    test effc mods in rocket fuel design on solid fuel part
    test beacons removed there as well
    test allaizns car beacon setup vs the other way (power for collision checks test)
    Make/find a mod that sets 0.17's recipes to 0.16's, then test (Rusty is doing this YAY!)
    Triple channel RAM, vs Dual, Vs multiple dimms in 1 channel
    Is it better to optimize for passive provider sharing or forego sharing to achieve the shortest travel distance on bot builds?
    Testing copied designs with various memory frequencies to validate best megabase scaled design. At megabase scale, the game becomes more memory bound.
    Testing with a dummy megabase, so that you are not forced to clone the thing you want to test beyond the scale it would otherwise be at
    Does expanding the map far out have performance penalty? Even when those chunks are not doing anything?
    How expensive is setting requests by the circuit network.
    See if miners having only one ore tile changes their performance.
    Test train signals helping performance (potential due to checking the rail blocks)
    What functions do active/inactive but disabled inserters check?
    does overlapping vehicle map markers have additional render cost?
    Does a dual headed train traveling backwards (relative to the "front" locomotive which is determined when a train is created) have performance loss?
    Backed up miners don't consume power, do they still perform electric network calculations?
    MIGRATION OF MOD OR FACTORIO VERSION
    Test orientation flipped cargo-wagons within a train (half 0 half 0.5, interleave or in blocks)
    is it better to activate all clocked inserters in one tick or space them out?
    When clocking inserters should you pulse for 12 items or set to active to unload 96 items and sleep in between for longer.
    How much do clocked inserters help with bot travel?
    is it better to change a value of a signal via 1->2->1  or 0->1->0, since one keeps the signal, while the other deletes and recreates it all the time
    Test if rezipping a map so it actually gets compressed affects UPS of a benchmark of that map.
    Test if it's better to split circuit networks or have 1 huge one
    WIP green circuit builds (test-8)
    DONE -- WIP electric networks (test-13)
    Design something around pipes that empty periodically.
    Does a train having a rider increase UPS cost?
    Do bots contribute to additional collision checks when they're hovered over a moving car example.
    Is stationary fluid cheaper to update than moving fluid?
    What about empty pipes is cheap? Lack of fluid or lack of fluid filter?