refactor: trace_decoder::decoding

[0xaatif]

#275

Change of pace to do some simple cleanup:

• Most of the module consisted of static methods on ProcessedBlockTrace, I've made those functions now.
• Removed all the custom error logic, and just use anyhow. The same details will be exposed to end users.
  • We may have to change error formatting elsewhere to get the full error chain, as the new approach makes use of std::error::Error::source, but no changes should be required for errors which [panic] or are returned from main.
• Made the existing code less verbose in places

[iljakuklic]

There are some nice changes and some potentially mildly controversial ones. See some comments below, mostly to do with general Rust coding conventions and such.

Note that the change in indentation makes it harder to see which are the genuine changes introduced in this PR. Therefore, some of the comments may refer to stuff that only incidentally shows up in the diff.

[muursh]

@Nashtare you mentioned some edge cases from John?

[Nashtare]

@muursh They mostly revolve around revert cases (either full txn revert or some chile context reversion). There seem to be some issues with txn reversions in mainnet blocks as well (cf #475, #421), but not sure yet whether this can be handled without changes on the Jerigon side or not, still looking at them.

[0xaatif]

@Nashtare you mentioned some edge cases from John?

I'm aiming for no behavioural changes with this PR - have I inadvertently broken something?

[atanmarko]

Only few non blocking nitpicks