Convert live entries to init entries in lowest level of bucketlist

[ThomasBrady]

Description

Resolves #193

Changes BucketOutputIterator::put to convert LIVEENTRY to INITENTRY in the lowest level of LiveBuckets.

Checklist

• Reviewed the contributing document
• Rebased on top of master (no merge commits)
• Ran clang-format v8.0.0 (via make format or the Visual Studio extension)
• Compiles
• Ran all tests
• If change impacts performance, include supporting evidence per the performance document

[ThomasBrady]

@SirTyson can you PTAL?

[SirTyson]

Overall LGTM. Just one logic cleanup and I think a higher level test would be more helpful here vs. a low level one.

[SirTyson]

Sorry, I think I was a little unclear in my last comment. A little cleanup on the interface, and I think the test case still needs improving.

[SirTyson]

This doesn't actually test our fix. The problem is, in old bucketlist code that has long since been fixed, live entries where written to the bottom level. The current code doesn't exhibit this behavior. What you're testing here is if the current BucketList logic works, but not that your conversion code from the buggy BucketList -> good bucketlist actually works. What we should do is populate a BucketList, but specifically populate it with buggy contents such that the bottom level has live entries (see "live bucket tombstones expire at bottom level" for populating a bucketList without addBatch). Then, add empty batched to your BucketList until all entries have coallesced in the bottom bucket (see "hot archive bucket tombstones expire at bottom level" for this example). Finally, scan the bottom bucket and make sure they're all init.

[SirTyson]

Test looks great! One issue that I didn't notice (until your test case, which was super helpful), we need to gate this feature on protocol 23. Currently, we gate on FIRST_PROTOCOL_SUPPORTING_INITENTRY_AND_METAENTRY, which is before p23. This means catchup will fail, as we'll change the bucket hashes for p22 for example. We need to add a new constexpr, like FIRST_PROTOCOL_CONVERTING_BOTTOM_LEVEL_LIVE_TO_INIT or something like that which is equal to 23. Sorry I didn't catch this before!

[ThomasBrady]

@SirTyson I had to rebase, can you re-enable automerge?