shred: speed up agave harness

[mjain-jump]

No description provided.

[Copilot]

Pull request overview

This PR targets harness runtime by reducing RocksDB degradation across repeated shred-parse inputs while keeping the same overall shred parsing / FEC recovery flow.

Changes:

• Add a per-thread input counter and periodically reopen the thread-local Blockstore to avoid accumulating RocksDB tombstones that slow reads over time.
• Keep the existing per-input purge_slots(..) reset for reused Blockstore instances, but force a full reopen every BLOCKSTORE_REOPEN_EVERY inputs.
• Bump the pinned firedancer-io/agave git revision across Agave/Solana git dependencies (and corresponding Cargo.lock entries).

Reviewed changes

Copilot reviewed 2 out of 3 changed files in this pull request and generated no comments.

File	Description
src/shred.rs	Periodically recreates the ephemeral thread-local blockstore to avoid RocksDB slowdowns from accumulated delete markers.
Cargo.toml	Updates the Agave git rev for multiple dependencies.
Cargo.lock	Updates locked sources to match the new Agave git rev.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.