fix!(share/byzantine): use any available axis for befp nmt proofs construction

[walldiss]

We should not restrict the BEFP constructor to collect proof only for orthogonal axes to ErrByzantine. This PR enables the constructor to attempt building both Row and Col proofs, irrespective of the ErrByzantine axis type. Additionally, it prevents the BEFP constructor from requesting proofs pr shars from the network by granting access exclusively to the local blockstore. There should be sufficient proofs and shares in the local blockstore at the time BEFP is detected.

Breaks befp message, by introducing proofAxis field, which is not supported by older befp subscribers.

Depends on

• feat(share/eds): Store nmt nodes on failed reconstruction #3305 to have enough data in blockstore.
• fix: deepcopy filliing nil data with zeroes rsmt2d#310 to provide proper coordinates of verified shares

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 67.27273% with 36 lines in your changes are missing coverage. Please review.

Project coverage is 45.01%. Comparing base (2469e7a) to head (eaa288b).
Report is 30 commits behind head on main.

Files	Patch %	Lines
share/eds/byzantine/pb/share.pb.go	0.00%	22 Missing and 2 partials ⚠️
share/eds/byzantine/share_proof.go	86.44%	7 Missing and 1 partial ⚠️
share/eds/byzantine/byzantine.go	73.33%	2 Missing and 2 partials ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3306      +/-   ##
==========================================
+ Coverage   44.83%   45.01%   +0.17%     
==========================================
  Files         265      272       +7     
  Lines       14620    15162     +542     
==========================================
+ Hits         6555     6825     +270     
- Misses       7313     7569     +256     
- Partials      752      768      +16     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[Wondertan]

Needs review from the author - @vgonkivs

[Wondertan]

Great find, seriously. It should have become obvious to us after we realized that the server is responsible for selecting the proof when we sample and it did become👍🏻

[vgonkivs]

--- FAIL: TestBEFP_ValidateOutOfOrderShares (0.00s)
    bad_encoding_test.go:241: 
        	Error Trace:	/home/runner/work/celestia-node/celestia-node/share/eds/byzantine/bad_encoding_test.go:241
        	Error:      	Received unexpected error:
        	            	incorrect share received
        	Test:       	TestBEFP_ValidateOutOfOrderShares
FAIL
coverage: 13.2% of statements

[walldiss]

I had to remove concurrent fetching of ShareWithProof code to resolve flakiness of unit tests. Also looks much simpler now. Performance downside is not important, as we collect shares from disk and it is still fast enough being sequential instead of concurrent.

[oblique]

I'm evaluating your changes to port them into Rust's celestia-types.

Correct me if I'm wrong but isn't this the same as the following?

	switch s.Axis {
	case rsmt2d.Row:
		rootHash = dah.RowRoots[axisIdx]
	case rsmt2d.Col:
		rootHash = dah.ColumnRoots[axisIdx]