Parity db migration

[i1i1]

Fixes #567

This pr migrates to parity db from rocksdb. I decided to migrate only piece index hash db (copying data over to parity db) and just remove all commitments (as we can regenerate them, and it shouldn't take much time).

Code contributor checklist:

• I have read, understood and followed contributing guide

[i1i1]

Here is the issue related to the release of parity-db

[i1i1]

Benchmarked changes, and it seems that performance has decreased:

Before:

$ cargo run --release -p subspace-farmer -- --base-path ... bench --plot-size 3G --write-pieces-size 6G
Finished benchmarking.

3.13G allocated for farming
2.79G actual space pledged (which is 89.25%)
344.44M of overhead (which is 10.75%)
1m 26s plotting time
34.69M/s average plotting throughput
Recommitment took 1.239412406s

After:

$ cargo run --release -p subspace-farmer -- --base-path ... bench --plot-size 3G --write-pieces-size 6G
Finished benchmarking.

3.20G allocated for farming
2.79G actual space pledged (which is 87.31%)
415.85M of overhead (which is 12.69%)
2m 24s plotting time
20.72M/s average plotting throughput
Recommitment took 1.071120161s

[nazar-pc]

Hm... looks like it is using a bit more space, slower to write, but faster to read 🤔

Overall makes sense to me.

[i1i1]

@nazar-pc reverted last commit with that abstraction for commitments database. Despite that history is the same

[nazar-pc]

I'd wait for upstream PR to be accepted though or at least address their concerns as we can't afford have stack overflows.