sync: cap pending Gloas column sidecars per peer

[satushh]

What type of PR is this?

feature

What does this PR do? Why is it needed?

Implements ethereum/consensus-specs#5199 (Gloas data column sidecar queueing).

When a Gloas data column sidecar arrives over gossip before its beacon block has been seen, it's queued for deferred validation. Today in develop, that queue admits sidecars on first-come-first-served basis and stores only one per (root, index), which means a single peer can occupy every queue slot and a malicious peer's invalid sidecar can hide behind another peer's good one.

Changes

1. Per-peer queue caps to prevent any single peer from monopolizing queue space:
   - maxPendingGloasColumnsPerPeer = 128 (total queued columns per peer).
   - maxPendingGloasRootsPerPeer = 2 distinct block roots per peer (could be increased if it is too strict).
   - Counters tracked on Service (pendingGloasPeerColumnCounts, pendingGloasPeerRootCounts) and released on flush/prune.
2. Multi-peer sidecar retention. pendingColumnEntry.sidecar/peer → pendingColumnEntry.sidecars map[peer.ID]*DataColumnSidecarGloas. The queue now retains every (peer, sidecar) pair received for a (root, index), deduplicated per peer.
3. Per-peer attribution on deferred validation. Every peer that forwarded a sidecar for an unseen-block (root, index) is retained, so when the block arrives and verification runs, each peer's offering is verified individually and bad peers can be downscored by ID.

Acknowledgements

• I have read CONTRIBUTING.md.
• I have included a uniquely named changelog fragment file.
• I have added a description with sufficient context for reviewers to understand this PR.
• I have tested that my changes work as expected and I added a testing plan to the PR description (if applicable).