Support WAN SciTag markers for remote HTTP(S) TPC transfers

[ShawnMcKee]

Summary

This PR implements the agreed Patch 1-3 scope for WAN SciTag marker support.

Patch 1: Pool marker protocol allowlist

• Add remotehttpdatatransfer and remotehttpsdatatransfer to TransferLifeCycle.needMarker(...).
• Keep WAN protocol scope as: xrootd, http, https, remotehttpdatatransfer, remotehttpsdatatransfer.

Patch 2: WebDAV TPC transfer-tag plumbing for remote HTTP(S)

• Capture SciTag carriers from COPY request in CopyFilter (SciTag, TransferHeaderSciTag).
• Pass selected transfer tag into RemoteTransferHandler.acceptRequest(...) and RemoteTransfer.
• Set transfer tag on both remote HTTP and remote HTTPS protocol-info objects.
• Add serialized transfer-tag storage and getter support to RemoteHttpDataTransferProtocolInfo (used by HTTPS subclass as well).

Patch 3: Access-log observability

• Add tpc.scitag and tpc.transferheaderscitag fields in WebDAV LoggingHandler access logs.

Validation Evidence

Validated on umfs19 with dcdum02 TPC traffic:

• pool protocol observed: remotehttpsdatatransfer
• exact tag matches confirmed (pool transferTag equals door tpc.scitag) for multiple pnfsids, including:
  • 00004C3CC39D92B94DF2B02B4A870AF09DA8 (144)
  • 00004677DBAD412943CAB37A32E479D3C103 (143)
  • 0000E1813ED9C11149D19137F03ABE1CC628 (144)
  • 00000FDBDB1E8DEA4FC7B2FD25FDC4C2361D (144)

Out of Scope

• No gftp, dcap, or nfs4 protocol expansion in this PR.
• No optional GSIFTP extension in this PR.

[kofemann]

Hi @ShawnMcKee, thanks for the contribution. I have some minor comments. The code looks good. Please annotate commits with Signed -off-by: tags.

[kofemann]

The extra null check is redundant, as initial value and setter alwast set "" for nulls

[ShawnMcKee]

Good catch. I removed the redundant null check in getTransferTag(). The value is normalized to "" by the setter and by deserialization fallback, so returning the field directly is sufficient. Included in commit 5b4768f.

[kofemann]

I wish there was a comment saying that "-" is a magic placeholder to indicate nulls.

[ShawnMcKee]

Added an explicit sentinel constant and comment to document the magic placeholder: MISSING_HEADER_VALUE = "-" now represents missing/blank SciTag header values. I also replaced the literal "-" checks with that constant for readability. Included in commit 5b4768f.

[kofemann]

It's good practice to have javadoc for public methods

[ShawnMcKee]

Agreed. I added Javadocs for the public accessor methods introduced for request attributes (getTpcSource, getTpcDestination, getTpcSciTag, and getTpcTransferHeaderSciTag). Included in commit 5b4768f.

[ShawnMcKee]

Note on validation with all three SciTag PRs (8039, 8040, 8042):

To test the full SciTag stack together, I created branch combo/8039-8040-8042.

Why this was needed:

• PR Fix TransferLifeCycle SciTag boundary and FQAN fallback mapping #8039 and PR Support WAN SciTag markers for remote HTTP(S) TPC transfers #8042 are on the master lineage.
• PR Firefly: send to flow destination and optional collector #8040 was merged into the 11.2 branch, not master.

So a direct master-only test branch would miss #8040. The combo branch starts from the 11.2 side (to include #8040) and layers the corresponding #8039/#8042 changes for integrated validation.

[kofemann]

Thanks for the update.I have merges #8040 into the master too. In general, please issue PR against master. We will do the backports.