[CI] SnapshotStressTestsIT must not close indices during partial snapshot

[DaveCTurner]

The partial snapshotter releases its index permits as soon as the snapshot has started, which frees up the restorer to close or delete these indices in order to restore them even while the snapshot is ongoing. Deleting an index while it's being (partially) snapshotted is fine, but closing them is forbidden:

elasticsearch/server/src/main/java/org/elasticsearch/cluster/metadata/MetadataIndexStateService.java

Lines 328 to 335 in 4076aa7

	Set<Index> snapshottingIndices = SnapshotsService.snapshottingIndices(currentState, indicesToClose);
	if (snapshottingIndices.isEmpty() == false) {
	throw new SnapshotInProgressException(
	"Cannot close indices that are being snapshotted: "
	+ snapshottingIndices
	+ ". Try again after snapshot finishes or cancel the currently running snapshot."
	);
	}

elasticsearch/server/src/main/java/org/elasticsearch/cluster/metadata/MetadataIndexStateService.java

Lines 868 to 882 in 4076aa7

	// Check if index closing conflicts with any running snapshots
	Set<Index> snapshottingIndices = SnapshotsService.snapshottingIndices(currentState, Set.of(index));
	if (snapshottingIndices.isEmpty() == false) {
	closingResults.put(
	result.getKey(),
	new IndexResult(
	result.getKey(),
	new IllegalStateException(
	"verification of shards before closing " + index + " succeeded but index is being snapshot in the meantime"
	)
	)
	);
	logger.debug("verification of shards before closing {} succeeded but index is being snapshot in the meantime", index);
	continue;
	}

We should make partial snapshotting and closing an index mutually exclusive.

[elasticsearchmachine]

Pinging @elastic/es-distributed (Team:Distributed)

[DaveCTurner]

Although this is a low-risk test-only bug, it's basically blocking attempts to reproduce #108907 which seems more serious, so I'm looking into this now.