perf(sdk): memoize history selectors and amount conversion factors (#32)

[RaymondAbiola]

Closes #366

[drips-wave]

@RaymondAbiola Great news! 🎉 Based on an automated assessment of this PR, the linked Wave issue(s) no longer count against your application limits.

You can now already apply to more issues while waiting for a review of this PR. Keep up the great work! 🚀

Learn more about application limits

[Jaydbrown]

Not mergeable as-is — two real problems, both blocking:

1. Removes strictness flags from all three tsconfig files repo-wide (tsconfig.json, tsconfig.rollup.json, packages/react/tsconfig.json):

   -    "noImplicitAny":           true,
   -    "useUnknownInCatchVariables": true,

   This is a real regression in the codebase's type-safety guarantees, applying to every file in the SDK and the React package, not just the selector-memoization work this PR is about. This almost certainly exists to let the new code type-check without properly typing something (implicit anys or untyped catch bindings) rather than being an intentional, discussed policy change. Please restore both flags and fix whatever type errors that surfaces in the new code instead.

2. This branch is stale and would silently revert an already-merged fix. src/dashboard/transaction-history.ts's toIso8601/SPACE_SEPARATED_DATETIME helper (added by PR #406, "normalize space-separated indexer timestamps to ISO 8601 for Safari") is removed by this diff, along with its regression test in src/tests/transaction-history-crash.test.ts. GitHub shows this as a clean, conflict-free diff since the textual regions don't overlap — but merging as-is would reintroduce the Safari timestamp bug #406 just fixed. Please rebase onto current main and drop this reversion; the actual memoization work in transaction-history.ts's selectors doesn't need to touch asTimestamp/toIso8601 at all.

The core selector-memoization approach and the utils.ts amount-conversion caching look reasonable on a skim, but I'd like to see this rebased and with the tsconfig changes reverted before a full re-review.

[RaymondAbiola]

Ok thanks for your review. I will fix it.

[RaymondAbiola]

@Jaydbrown Thanks for the review — rebased onto current main, both blockers are resolved.

On the tsconfig flags: the branch never actually touched any tsconfig file. noImplicitAny and useUnknownInCatchVariables were added to main after this branch forked (base was faefc1b, main is now 944baf3), so the diff rendered them as removals against a stale base. Same root cause as the toIso8601 reversion you spotted — the rebase fixed both. All three tsconfigs and the #406 Safari timestamp fix now show zero diff against main.

On the rest: the rebase showed the optimizations here had already landed:

• POW10 lookup table — feat: improve performance in utils.ts and streams.ts with caching and precomputation #405
• bigintSafeStringify lazy cloning — Feat/381 detailed sdk feature #414
• Selector filter/sort memoization — perf(dashboard): O(n log k) pagination for selectVisibleTransactions #408

I resolved every conflict in favor of main and dropped my implementation. The WeakMap + O(n log k) top-k approach in #408 is better than what I had, and keeping mine would have been a regression. The docs changes were redundant with the "Selector memoisation" section already there.

I also verified the signature string can't collide (the first three positions are enum-constrained and space-free) and pinned that down with a test, since a search term containing spaces is the obvious way it could break.