Deprecate bz2/gzip/none/tar snapshot compression types

[steviez]

Problem

For our use cases, the compression types being removed are inferior and have caused us support burden.

Summary of Changes

These options are now disallowed on the command line for solana-validator and solana-ledger-tool, which effectively means no more snapshots will be created with this types in normal usecases. However, support for reading the deprecated types is still in place.

This PR is a follow-on to #33442 (which I will be closing); that PR ripped several things out whereas this PR has a more gradual deprecation approach in mind

[codecov]

Codecov Report

Merging #33484 (3c4d652) into master (3508b7d) will increase coverage by 0.0%.
Report is 1 commits behind head on master.
The diff coverage is 100.0%.

@@           Coverage Diff           @@
##           master   #33484   +/-   ##
=======================================
  Coverage    81.7%    81.7%           
=======================================
  Files         802      802           
  Lines      217801   217792    -9     
=======================================
+ Hits       178114   178130   +16     
+ Misses      39687    39662   -25     

[steviez]

As a quick sanity check, here is ouptut running with a now deprecated value:

$ cargo run -- create-snapshot 10 --ledger ~/ledger --snapshot-archive-format none
    Finished dev [unoptimized + debuginfo] target(s) in 2.28s
     Running `.../solana/target/debug/solana-ledger-tool create-snapshot 10 --ledger ~/ledger --snapshot-archive-format none`
error: 'none' isn't a valid value for '--snapshot-archive-format <ARCHIVE_TYPE>'
	[possible values: lz4, zstd]

It is worth calling out that it will be a hard error for using old type. If we want to be even gentler, we could still accepts none, bzip2, ... and just translate those to zstd under the hood + emit a warning.

[brooksprumo]

Works for me. Can you confirm this does the right/expecting thing on solana-validator and solana-ledger-tool? I'm wondering what the error messages are if specifying "none", since that'll affect some current node operators when the eventually upgrade.

Edit: Hah, you beat me to it!

[brooksprumo]

It is worth calling out that it will be a hard error for using old type. If we want to be even gentler, we could still accepts none, bzip2, ... and just translate those to zstd under the hood + emit a warning.

I'd rather hard error, personally. I'm not sure there's a valid reason to use "none" anymore, given how much faster "zstd" is. But this is not a strongly held opinion.

Minimally, we'll want this documented in release nodes/change logs.

[brooksprumo]

Lgtm. I'd like to have mvines review as well before merging.