## Summary

This PR upgrades the `github.com/prometheus/prometheus` dependency from
v0.307.3 to v0.308.0.

## Changes Made

#### Fixed `RemoteWriteProtoMsg` unrecognized

In prometheus/prometheus v0.308.0, the remote write protocol message
types were moved from `github.com/prometheus/prometheus/config` to
`github.com/prometheus/client_golang/exp/api/remote` as part of
[prometheus/prometheus#17197](prometheus/prometheus#17197).

#### Fixed `CreatedTimestamp` → `StartTimestamp`

In
[prometheus/prometheus#17411](prometheus/prometheus#17411),
the Remote Write 2.0 spec was updated to 2.0-rc.4. The
`CreatedTimestamp` field was renamed to `StartTimestamp` and moved from
`TimeSeries` to individual `Sample` and `Histogram` messages.

#### Fixed not enough arguments in`api_v1.NewAPI` 

The `api_v1.NewAPI` function signature changed due to
[prometheus/prometheus#17470](prometheus/prometheus#17470)
- Added `appendMetadata bool` parameter for remote write metadata
handling. Added the new parameter `appendMetadata = false` -
prometheusreceiver scrapes targets, doesn't receive remote write
requests

## Remaining Issues

### `EnableNativeHistogramsIngestion` removed from `scrape.Options`

There's one change in this upgrade that I'd love to get the community's
input on before finalizing.

In
[prometheus/prometheus#17315](prometheus/prometheus#17315),
Prometheus removed the `EnableNativeHistogramsIngestion` field from
`scrape.Options`. Native histogram ingestion is now controlled through
the Prometheus scrape config option `scrape_native_histograms` rather
than programmatically.

This creates an interesting situation for us. Currently, we have the
feature gate `receiver.prometheusreceiver.EnableNativeHistograms` that
controls whether native histograms are converted to OTel exponential
histograms. But with this Prometheus change, the *ingestion* of native
histograms during scraping is now a separate concern controlled by
Prometheus config.

I see two paths forward here. We could automatically inject
`scrape_native_histograms: true` into the scrape config whenever our
feature gate is enabled, preserving the current behavior where enabling
the feature gate is all users need to do. Alternatively, we could simply
remove the line and require users to configure
`scrape_native_histograms: true` in their Prometheus scrape config
themselves, which is more explicit but adds an extra configuration step.

What do you think? I'm also wondering if this constitutes a breaking
change that should be called out in the changelog. Would appreciate any
thoughts!

---------

Signed-off-by: Arthur Silva Sens <arthursens2005@gmail.com>