Cleanup error type mapping in serde_snapshot

[svenski123]

Problem

Excessive explicit error type mapping is unclear

Summary of Changes

Rework error types to allow use of '?' operator without explicit map_err() calls

[codecov]

Codecov Report

Merging #10580 into master will increase coverage by 0.0%.
The diff coverage is 45.4%.

@@           Coverage Diff           @@
##           master   #10580   +/-   ##
=======================================
  Coverage    81.7%    81.8%           
=======================================
  Files         301      301           
  Lines       70634    70614   -20     
=======================================
  Hits        57775    57775           
+ Misses      12859    12839   -20     

[ryoqun]

Thanks for working on general clean-ups!

[ryoqun]

This ill-taste continue was a work-around introduced at: https://github.com/solana-labs/solana/pull/7281/files#diff-2099c5256db4eb5975c8834af38f6456R503

With good confidence, we no longer need this. This isn't no longer warning. It's safe to cause error and abort now as this is really bad condition now. Could you remove continue and adjust the new code so that the work-around in that dated pr will be effectively reverted?

[ryoqun]

(For the record: this workaround got deprecated at #8339)

[ryoqun]

Yay! this reads a lot better. :)

[ryoqun]

LGTM

Flawless and well-thought cleaning PR ever! Thanks!

[ryoqun]

@svenski123 Admitted, error handling is rough areas in our codebase. Thanks for polishing! You might find these interesting as a further direction towards better error handling:

• It seems that there are bunch of places doing format!() or .to_string() needlessly when casting to std::io::ErrorKind::Other.
• It seems that there are bunch of places spelling out Box<bincode::ErrorKind> which can be just said bincode::Error as you did in this pr.
• Validator::new is chaos and it's doing process::exit because of the unproper use of ::new() -> Self convention (no Result<> here). It just needs more idiomatic rusty instantiation and break up the code...

  solana/core/src/validator.rs

  Line 225 in 50c93d4

  process::exit(1);

[svenski123]

Regarding error types, I think something like https://docs.rs/anyhow/1.0.31/anyhow/ would make sense in a number of places.