op-batcher: merge the two tick loops and delete the clearStateRequested CAS mailbox

[philippecamacho]

The two tick loops each poll getSyncStatus and coordinate through the clearStateRequested CAS mailbox because neither owns all the state. Merging them into one loop (keeping both tickers, so cadence and observable behavior are unchanged) lets reset() call clearState directly and deletes the mailbox protocol entirely.

Original discussion: #459 (comment)

[jjeangal]

When a reset triggers clearState and L1 safe-origin fetching is failing, is it acceptable for both block-queueing and streamer-draining to pause until it succeeds (or shutdown), rather than only draining pausing as today? For example, if clearState were to block indefinitely, the block-queueing and streamer-draining steps would block each other, since they now run on the same goroutine.
CC. @philippecamacho

[philippecamacho]

When a reset triggers clearState and L1 safe-origin fetching is failing, is it acceptable for both block-queueing and streamer-draining to pause until it succeeds (or shutdown), rather than only draining pausing as today? For example, if clearState were to block indefinitely, the block-queueing and streamer-draining steps would block each other, since they now run on the same goroutine. CC. @philippecamacho

I think it is ok as if L1 safe-origin fetching is failing then queueing loop call to getSyncStatus return an error as well.

[jjeangal]

Revisiting the approach here based on Sneh's review on #521. Rather than merging the two loops onto one goroutine, we're going with a mutex. Guard the shared streamer position with a dedicated espressoStreamerMutex, inline and consistent with channelMgrMutex. That's a light, no-behavior-change step (mailbox stays), landing in a separate PR.

Deleting the mailbox for true parallel queueing/loading is deferred to the streamer redesign: it needs streamer-level guarantees (atomic peek-and-advance, a bounded validity-check call) that the current streamer doesn't provide. Repurposing this issue to track that mailbox removal as part of the streamer work, plus Sneh's two follow-ups (over-clearState on non-advancing errors; Espresso tx-size-limit handling).