1.21.1-NF: vault no longer cancels fast run

[erutobusiness]

Vaulting over obstacles cancels vanilla sprint due to horizontal block collision in the first couple ticks of the vault animation. Since FastRun.canContinue() requires player.isSprinting(), this caused fast run to stop mid-vault, leaving the player walking after the vault completes.

The fix registers a sprint cancel marker in Vault.onStartInLocalClient so that the LivingEntityMixin setSprinting hook blocks vanilla's cancellation for the duration of the vault. This is the same pattern already used by RideZipline.

[alRex-U]

If I remember correctly addMarkerCancellingSprint will stops sprint always?
Anyway currently I'm working on v4 update on v1.19.2 branch, it overwrite most of basic system.
Please let me put off merging

[erutobusiness]

Thanks for looking at it, and no rush at all on merging!

You're right that in RideZipline addMarkerCancellingSprint is used to keep sprint off — but that's because it's paired with an explicit setSprinting(false) right before the marker (RideZipline#onStartInLocalClient). The marker itself doesn't force sprint off: in LivingEntityMixin.onSetSprinting it only cancels the setSprinting() call (ci.cancel() at HEAD), so it just freezes whatever the current sprint state is.

In this PR I register the marker without calling setSprinting(false) first, so during a vault — where the player is already sprinting for fast run — it holds the sprint on and blocks vanilla's collision-triggered setSprinting(false) for the couple of ticks the vault lasts. So it's the same enforcer, just relied on to hold the current state instead of to suppress sprint.

If you'd rather not reuse the marker in the opposite direction like this, I'm open to a different approach — just let me know what fits ParCool best.

And totally fine to put off merging. Since the v4 rewrite on 1.19.2 reworks most of the basic system (and will flow up into this branch), please take your time. Once v4 is ready I'm happy to rebase or reimplement this on top of the new system, so just tell me if the fast-run-after-vault behavior still needs attention then.