fix: add network column to indices

[morph-dev]

What was wrong?

Some of our queries were very slow because our db indices weren't setup correctly.
For example, the TOTAL_DATA_SIZE_QUERY_DB query took 20-25 seconds on 4GB database.

How was it fixed?

I added network column to the indices. Based on Kolby's measurements on the same db, we noticed following improvements:

query	before (sec)	after (sec)
TOTAL_DATA_SIZE_QUERY_DB	23.087	0.008
XOR_FIND_FARTHEST_QUERY_NETWORK	6.154	0.017

To-Do

• Clean up commit history and use conventional commits.

[KolbyML]

Amazing find 🚀.

[KolbyML]

Looking at our sql queries should we keep the index for content_id_long?

[morph-dev]

I looked more online and again into our queries and made few changes.

Looking at our sql queries should we keep the index for content_id_long?

I confirmed both by manual inspection and by this post that index is not needed for the primary key or unique column (and it might hurt) because they are automatically created.

So I also dropped the index for primary key from lc_update table.

I also figured out that:

• content_id_short_idx_2 should include content_id_long
  • alternative is to put LIMIT 1 to the XOR_FIND_FARTHEST_QUERY_NETWORK query. Currently we only read first item, but potentially we could read more and delete more in a batch in which case the index is useful
• the (network, content_id_long_idx) was not needed. Instead I simplified the TOTAL_ENTRY_COUNT_QUERY_NETWORK query so it can rely on network_idx
• the we needed index for content_key because of the PAGINATE_QUERY_DB query (this one shouldn't have network column)

Please take another look

[njgheorghita]

🚀