[GOT-66] Create stress tests to understand how read-query load affects syncing time

[jsign]

We should try to understand better how read-query load affects background work regarding syncing chains.

Technically speaking, we use WAL mode in SQLite, so readers shouldn't block writers, compared to the default journal mode where this isn't the case. Using WAL mode was done intentionally exactly for this reason.

This can help us understand better the limits of the validator in particular VM setups (e.g: the one we use in the cloud).

In any case, reads are easy to scale since they can be executed in "replica" nodes without interfering with writes. That's a more complex setup since it requires SQLite replication (we can use excellent projects out there) and other processes running, but is totally doable. But, it might be good to understand that limit better and know if we're getting closer to working on it.

_GOT-66