Reuse metadata and protocol entries while retrieving the active files

[findinpath]

Description

This is an incremental improvement in the direction set by #18916

While retrieving the metadata & protocol entries from a multi-part checkpoint file, stop scanning the checkpoint files as soon as the metadata & protocol entries are actually found.

Reuse metadata and protocol entries while retrieving the active files

The metadata & protocol entries are already read (and saved) once
when retrieving the table handle.
Reuse this information while retrieving the active files for the table.

Additional context and related issues

Release notes

(x) This is not user-visible or is docs only, and no release notes are required.
( ) Release notes are required. Please propose a release note for me.
( ) Release notes are required, with the following suggested text:

# Delta Lake
* Avoid redundant reading of the checkpoint files. ({issue}`issuenumber`)

[findepi]

Good find

[findepi]

The .*.parquet is pretty broad.
For example, checkpoint files would also match the pattern.

Maybe instead of this,
let's turn dataFilePattern into an ordinary catch all pattern \Q table_directory \E / ( key=value /)? [^/]+

[findinpath]

Modified the pattern to .*?/(?<partition>key=[^/]*/)?[^/]+ and placed it only after matching against metadata files.

[findepi]

remove

if (entryTypes.contains(ADD)) {
            metadataAndProtocol = Optional.of(getCheckpointMetadataAndProtocolEntries(
                    session,
                    checkpointSchemaManager,
                    typeManager,
                    fileSystem,
                    stats,
                    checkpoint));

from io.trino.plugin.deltalake.transactionlog.TableSnapshot#getCheckpointTransactionLogEntries

it's now dead code

[findinpath]

Not sure about this:

trino/plugin/trino-delta-lake/src/main/java/io/trino/plugin/deltalake/transactionlog/checkpoint/CheckpointWriterManager.java

Lines 116 to 123 in b7d7ba8

	snapshot.getCheckpointTransactionLogEntries(
	session,
	ImmutableSet.of(PROTOCOL, TRANSACTION, ADD, REMOVE, COMMIT),
	checkpointSchemaManager,
	typeManager,
	fileSystem,
	fileFormatDataSourceStats)
	.forEach(checkpointBuilder::addLogEntry);

[findepi]

the linked code place already reads metadata entry.
let it read protocol as well and pass both

[findinpath]

@findepi pls test this PR with secrets.

[findepi]

/test-with-secrets sha=f308b77d8670341f48b98c4503b86baea449575c

[findepi]

🚀