consensus: use batch for genesis queries

[pro-wh]

this kinda sucks

• db slower ~13s -> ~17s on eden genesis file
• can no longer print out queries into a migration-like file

also some of the ON CONFLICT is not exactly preserved, so might be some hidden bugs in network upgrades

measurements, very informal, only one run each

reuse
{"analyzer":"consensus","caller":"genesis.go:52","count":355437,"height":"genesis","level":"info","module":"analysis_service","msg":"generated genesis queries","ts":"2024-02-08T01:42:57.758442953Z"}
{"analyzer":"consensus","caller":"consensus.go:204","level":"info","module":"analysis_service","msg":"genesis document queries generated","process_duration":"32.732929903s","ts":"2024-02-08T01:42:57.758504654Z"}
{"analyzer":"consensus","caller":"consensus.go:228","level":"info","module":"analysis_service","msg":"genesis document processed","send_batch_duration":"17.026809672s","ts":"2024-02-08T01:43:14.785349655Z"}

control
{"analyzer":"consensus","caller":"genesis.go:56","count":369,"height":"genesis","level":"info","module":"analysis_service","msg":"generated genesis queries","ts":"2024-02-08T02:30:31.547038915Z"}
{"analyzer":"consensus","caller":"consensus.go:204","level":"info","module":"analysis_service","msg":"genesis document queries generated","process_duration":"43.796619503s","ts":"2024-02-08T02:30:31.547102908Z"}
{"analyzer":"consensus","caller":"consensus.go:229","level":"info","module":"analysis_service","msg":"genesis document processed","send_batch_duration":"13.060895665s","ts":"2024-02-08T02:30:44.608061096Z"}

[pro-wh]

oh and hide whitespace on the diff due to some indentation decrease

[mitjat]

Huh, I see the dilemma.

• The 13->17s does suck for local development.
• The dumpable queries ... I don't care for particularly. They can be post-processed pretty easily to show in a readable format, IMO. Sth like output the query template only if it differs from the previous query, otherwise only output the args. Not that we should implement that now; it's just that whoever needs to debug genesis queries that deeply (I haven't, in recent memory) can hack together some insight easily enough.
• The code is so much cleaner, and more resilient to future DB changes.

I vote to check this in. I can wait the extra 4 seconds (times 30 iterations) next time I'm working on something that requires genesis processing.

And it can be sped up if needed:

• The queries are absolutely dominated by account balances (~280k) and staking delegations (~70); there's <10k other rows. So we only really need to speed the first two, or even just the balances.
• The data types involved are simple, so using the COPY protocol like so should be simple too.
  • The caveat is that COPY cannot do upserts. We do need upserts because we apply the genesis after fast sync. So you'd need to COPY into a temp table, and then do a INSERT FROM mytemptable INTO accounts ON CONFLICT ....
  • The exact implementation is a little messy too, but it's not super bad. We'd need to either pass the db connection into genesis.go (inversion of control from what we have now; I'm not a fan), or Process() in genesis.go would need to return a triple: the [][]interface{} for balances, the [][]interface{} for delegations, and the batch. The caller would then have to make both CopyFrom() calls plus exec the batch (which would contain commands for upserting from the temp tables).
• Conveniently, those are also the only two queries you needed to newly add into queries.go. (That's mostly irrelevant, except emotionally :))

Whether you want to take a stab at that is up to you. Even if not, I think this PR is a step in the right direction. Nice cleanup, thank you.

[pro-wh]

let's go with how this is now