[Gitflow] Merge master into develop

To ensure we have up-to-date TS support, e.g. `satisfies` cannot be
transpiled previously.

See https://github.com/alangpierce/sucrase/blob/main/CHANGELOG.md

feat(utils): Add function to extract relevant component name

Some small fixes to integrations & rewriting some "forgotten" ones to
functional syntax.

This scope lives in module scope and is applied to _all_ events.

Please review this carefully, as it is important that data is correctly
applied etc. There should be a decent amount of tests covering all of
this, but just to make sure. This was mostly ported/extracted from
node-experimental.

Refactors deno integrations to functional syntax.

To align with OTEL spans.

This also deprecates `span.finish()`, and also makes sure that `end()`
accepts both ms as well as s timestmaps (to align this with OTEL in the
future, which accepts ms only).

---------

Co-authored-by: Yagiz Nizipli <yagiz@nizipli.com>

Also ensure to actually export `getGlobalScope()` everywhere as well.

The isolation scope, currently, lies on the hub, and is applied to all
events.

…tyle (#9936)

Slowly getting there...

@lforst noticed that currently `startSpan()` and `startSpanManual` do
not fork the scope, which is also a slightly different behavior than it
is in OTEL.

This adjusts this to better align, so that we always fork a scope. We
also always leave the span on the forked scope, even after it was
finished.

In a follow up, we can thus also get rid of the `finish` callback arg
for `startSpanManual`, as users can/should simply call `span.end()`
themselves then.

…e-on-spans

feat(react): Send component name on spans

One of the PRs scoped from
#9855

Sends component names on the databag of interaction spans

One of the PRs scoped from
#9855

Sends component names on Replay UI breadcrumbs so they can be ingested
and indexed. This will allow for searching for Replays by component name
in the future.

---------

Co-authored-by: Yagiz Nizipli <yagiz@nizipli.com>

One of the PRs scoped from
#9855

Sends component names on UI event breadcrumbs

---------

Co-authored-by: Abhijeet Prasad <aprasad@sentry.io>

Adds https://biomejs.dev/linter/rules/no-unused-imports/

## Implements

- [x] 10 Second Bucketing: SDKs are required to bucket into 10 second
intervals (rollup in seconds) which is the current lower bound of metric
accuracy.
- [x] Flush Shift: SDKs are required to shift the flush interval by
random() * rollup_in_seconds. That shift is determined once per startup
to create jittering.
- [ ] Force flush: an SDK is required to perform force flushing ahead of
scheduled time if the memory pressure is too high. There is no rule for
this other than that SDKs should be tracking abstract aggregation
complexity (eg: a counter only carries a single float, whereas a
distribution is a float per emission).

## Caveats 

- Force flush requires Node.js 14+ support (FinalizationRegistry). I
recommend leaving it after v8 release to make the implementation a lot
easier.

---------

Co-authored-by: Abhijeet Prasad <aprasad@sentry.io>

I think this was just not updated... I aligned it with the webpack size
limit for the "full" bundle.

…t for CSP (#9969)

This PR adds options to the `sentryHandle` request handler that allows
control over CSP-relevant aspects for the `<script>` that the request
handler injects into the page. Previously, the injected script was
blocked by browsers if CSP was enabled, due to inline `<script>`s not
being allowed without a nonce or hash.

Now users can specify a nonce or disable the script injection

…anManual`, `trace` (#9958)

Our previous implementation of `startSpan` et.al. assumed
that because we were "cloning" the original promise, unhandled promise
rejections would still bubble up to the global `onunhandledrejection`
handler so that Sentry would catch the error.

However, we tried investigating this and found out that our "cloning"
mechanism didn't work correctly and because we already added a rejection
handler, the promise rejection would _not always_ make it to the global
handler.

After adding multiple integration tests, I further narrowed the buggy
behaviour down to a rather special case: The `startSpan` call is not `await`ed
which for some reason will lead to the SDK not catching the error.
Unless, we apply the fix in this PR.

This patch removes this cloning mechanism in favour of directly attaching a
rejection handler to the promise. In this handler, we rethrow the error
which should trigger the global handlers.