Improve range filter performance

[mattsse]

Describe the feature

Currently, the range filter fetches the entire block one by one:

reth/crates/rpc/rpc/src/eth/filter.rs

Line 316 in 041b8d3

if let Some(block) = self.block_by_number(block_number).await? {

this is suboptimal because we only need the body if the bloom matches

we do however need to read all the headers, this should be done first via the headers_range API which should be fast and then only fetch the full body via cache

TODO

• read header range (use some reasonable chunks so that we're not reading the entire range)
• only read full body if bloom is a match (reading the full block could likely be further optimized in the caching service if we already have already have the header so we only need to read the body)

Additional context

No response

[jerusan]

I can work on this

[mattsse]

hey @jerusan I already took this on in #2635

if you want to contribute, there are a few good first issues, like #2624