Cache whether a Table is a scheduler table, avoiding fetching the schema

[Centril]

Description of Changes

Fixes #2021.

Does 2 things:

1. adds fn update_mut_tx to traits.rs
2. adds Table::is_scheduler so datastore_{insert, update}_bsatn don't have to fetch the schema.

Perf before:

for i in {1..5}; do spacetime call basics update_positions_by_collect -s local; done

2025-01-17T20:57:06.015709Z  INFO: : Timing span "update_positions_by_collect": 615.027471ms
2025-01-17T20:57:07.248089Z  INFO: : Timing span "update_positions_by_collect": 618.673753ms
2025-01-17T20:57:08.473255Z  INFO: : Timing span "update_positions_by_collect": 607.102709ms
2025-01-17T20:57:09.581249Z  INFO: : Timing span "update_positions_by_collect": 608.032989ms
2025-01-17T20:57:10.787973Z  INFO: : Timing span "update_positions_by_collect": 606.664556ms

Perf after:

for i in {1..5}; do spacetime call basics update_positions_by_collect -s local; done

2025-01-17T22:52:31.389501Z  INFO: : Timing span "update_positions_by_collect": 569.702228ms
2025-01-17T22:52:32.599270Z  INFO: : Timing span "update_positions_by_collect": 585.807116ms
2025-01-17T22:52:33.686554Z  INFO: : Timing span "update_positions_by_collect": 597.645714ms
2025-01-17T22:52:34.893818Z  INFO: : Timing span "update_positions_by_collect": 586.649907ms
2025-01-17T22:52:35.994934Z  INFO: : Timing span "update_positions_by_collect": 585.149919ms

This means we go from an average 610 ms to 585 ms, which is about an average 25 ms speedup.
This was the expected speedup based on % of samples in the update ABI PR flamegraph.

API and ABI breaking changes

None

Expected complexity level and risk

1, fairly simple and contained change.

Testing

TODO