#0: Cleanup/optimize some dynamic noc code

[tt-aho]

Ticket

Link to Github Issue

Problem description

Suggestions from @pgkeller on some code cleanup/reorg and some performance changes.
Specifically, tagging the L1 reads/writes with tt_l1_ptr instead of using NOC_READ/WRITE_REG, and moving the register reads up front during init to pipeline and hide latency.

What's changed

Removed array and switched to compile time raw address calculation to make the address derivation clearer.
Tagged L1 reads/writes as tt_l1_ptr.
Moved register reads up front.

Checklist

• All post commit CI passes https://github.com/tenstorrent/tt-metal/actions/runs/13702284436
• Blackhole Post commit CI passes (if applicable) https://github.com/tenstorrent/tt-metal/actions/runs/13702289134
• Model regression CI passes (if applicable)
• Device performance regression CI passes (if applicable)
• (For models and ops writers) Full new models tests CI passes (if applicable)
• New/Existing tests provide coverage for changes

[pgkeller]

add a comment and then all good