1. fix(state): Stop reading redundant blocks for every FindHashes and FindHeaders request

[teor2345]

Motivation

Zebra reads 160-500 redundant blocks for each FindHashes or FindHeaders request.
The FindHeaders requests regularly take 5-10 seconds to answer.

This is a security issue because remote nodes can send these requests.

Solution

• Find a range of heights, rather than a list of hashes
• Stop reading redundant blocks and calculating redundant hashes
• Remove redundant state queries
• Allow some inconsistencies that can happen during concurrent writes, but don't do concurrent queries yet
• Simplify some of the code

This PR contains a regression - the header queries used to be partially concurrent, now they block the rest of the state.
The next PR will implement full concurrency.

Review

This is an urgent PR to fix Zebra sync performance.

Reviewer Checklist

• Code implements Specs and Designs
• Tests for Expected Behaviour
• Tests for Errors

[codecov]

Codecov Report

Merging #4825 (e772314) into main (ed553a9) will decrease coverage by 0.06%.
The diff coverage is 70.80%.

@@            Coverage Diff             @@
##             main    #4825      +/-   ##
==========================================
- Coverage   78.88%   78.82%   -0.07%     
==========================================
  Files         305      305              
  Lines       38559    38627      +68     
==========================================
+ Hits        30418    30447      +29     
- Misses       8141     8180      +39     

[oxarbitrage]

looks good to me

[teor2345]

Thanks for the review!

All the tests passed except for lightwalletd, which is fixed by PR #4828.
So I'm going to admin-merge this.