Update NoteId and NoteCommitment computation

[Fumuran]

Summary

This PR separates note details from the full public note identifier: it introduces NoteDetailsCommitment (recipient + assets only), defines NoteId as the merged hash of the details commitment and metadata commitment (matching block/batch note-tree leaves), and renames kernel memory, assembly, and call sites so terminology matches that model end-to-end.

Changes

Rust (miden-protocol)

• Adds NoteDetailsCommitment and wires NoteHeader / Note so the serialized header stores details commitment + metadata, while NoteId is derived via NoteId::new(details_commitment, metadata).
• Updates NoteDetails to expose commitment() for the details hash (replacing the old overloaded id() semantics).
• Adjusts block/batch note trees, Merkle verification, transaction inputs/outputs, batch trackers, advice inputs, and tests to use details commitment vs NoteId consistently; removes the previous compute_note_commitment-style helpers where superseded.

Transaction kernel (MASM + memory.rs)

• Renames offsets (INPUT_NOTE_DETAILS_COMMITMENT_OFFSET, OUTPUT_NOTE_DETAILS_COMMITMENT_OFFSET) and procedures (compute_input_note_details_commitment, compute_output_note_details_commitment, set_input_note_details_commitment) so offset 0 of each note segment holds the details commitment; authentication uses NOTE_ID (merged leaf) and docs use EMPTY_WORD_OR_NOTE_ID for the input-notes commitment hash second limb where applicable.
• Aligns kernel memory diagrams and constant names with the layout above.

Part of: #1731

[Fumuran]

Work should be continued after #2874 is solved