Clean up genesis init helpers

[arnetheduck]

• create testnet/test genesis starting from phase0 and upgrade to the correct fork, like the spec suggests
• reduce mem usage and number of copies while upgrading
• add gloas to ncli_testnet generator
• speed up mock deposit processing by skipping bls ops for most state creations and caching them for the others (attestation pool tests go from minutes to seconds)
• consolidate the various ways to create deposits and states in the test suite to a single flow
• make sure deposit creation uses the correct fork version in tests

[github-actions]

Unit Test Results

       12 files  ±0    2 424 suites  ±0   46m 8s ⏱️ - 17m 8s
12 630 tests ±0  12 065 ✔️ ±0  565 💤 ±0  0 ❌ ±0 
63 536 runs  ±0  62 808 ✔️ ±0  728 💤 ±0  0 ❌ ±0 

Results for commit 761ed5c. ± Comparison against base commit d1e888d.

♻️ This comment has been updated with latest results.

[etan-status]

There's something about current/next fork fields that's different when doing sequential upgrade instead of a usual genesis (where both fields are the same)

[arnetheduck]

There's something about current/next fork fields that's different when doing sequential upgrade instead of a usual genesis (where both fields are the same)

hm, yeah, we had these checks that I don't know where they came from that the genesis fork has the same versions in current/prev version .. I think that's our own invention and not a spec requirement per se so it was removed in #7705 - hopefully I nabbed all of them..

[etan-status]

Maybe an option to check genesis states from hoodi etc which started post phase0

[arnetheduck]

https://github.com/eth-clients/hoodi/blob/main/parsed/parsedConsensusGenesis.json:

arnetheduck@praeceps:~/Downloads$ grep _version parsedConsensusGenesis.json 
    "previous_version": "0x40000910",
    "current_version": "0x50000910",

[tersec]

Is the commented-out code intentional?

[tersec]

Oh, it's in the current code too.