fix: initialize large-binary S3 bucket at startup to prevent NoSuchBucketException

[kunwp1]

What changes were proposed in this PR?

Previously, the texera-large-binaries S3 bucket was created lazily (only when a workflow first used a large-binary operator). If no large-binary had ever been used, the bucket would not exist. This caused a NoSuchBucketException warning every time WorkflowService.clearExecutionResources ran at workflow initialization, since it unconditionally attempted to delete objects from the bucket.

This fix mirrors the existing pattern used for the dataset bucket in FileService: LargeBinaryManager.initialize() is now called once during TexeraWebApplication startup, ensuring the bucket always exists before any workflow execution or cleanup is attempted. The redundant per-call createBucketIfNotExist that was previously inside LargeBinaryManager.create() has also been removed, as it is no longer needed.

Any related issues, documentation, discussions?

Fixes #4239

How was this PR tested?

Check if the error message disappears after running any workflow

Was this PR authored or co-authored using generative AI tooling?

Claude-4.6

[kunwp1]

I suggest @mengw15 to review this PR as he filed the issue.

[chenlica]

@mengw15 Please review it.

[mengw15]

LGTM

[chenlica]

@aicam Please do another review.