implement minimum delegation

[2501babe]

for the initial bpf port, we had to disregard the unactivated minimum delegation feature gate to preserve existing behavior. now that we can modify the program via feature gate directly, we are free to impl this

im calling this medium priority... like not in the next couple months, other things are more important, but ideally this year

[2501babe]

when rewriting split i noticed that the original version failed to enforce minimum delegation for activating stakes because it only cared if effective > 0. i fixed this. its impossible to write a test for it with a 1lamp minimum but it should be added when we do this. also scan the other processor functions for similar issues

we should test mimimum delegation by compiling a 1lamp version into program/tests/fixtures/ since we may need to update tests/code to catch up on regressions that may have crept in, we have been running with no minmum for awhile and im not sure if tests with no test_case had features on or off (or a mix) since this is sometimes done inconsistently in underlying test harnesses