Reduce memory used when saving Performance data for offline viewing

[hkarmoush]

Fixes #10010

Saving Performance data wrote the Perfetto trace as a JSON array of numbers. jsonEncode then does roughly two string writes per trace byte, so a ~50 MB trace becomes a ~184 MB file and a very large number of string fragments in the browser, which is what runs the tab out of memory.

This changes the export to:

• write the trace as a base64 string under a new traceBinaryBase64 key (about 1.33x the trace size instead of about 3.7x). toJson() keeps a ByteData view and the existing toEncodable does the base64 step, so "Review History" on disconnect still holds the trace without copying it.
• keep reading the old traceBinary list format, so existing exported files still load. Older DevTools versions opening a new file will see no trace rather than throwing a type error, because the key is different.
• fix ByteDataEncodeDecode to encode only the bytes in the view, not the whole backing buffer.
• allocate Uint8ListRingBuffer.merged once at its exact size instead of letting BytesBuilder round up to a power of two.
• show an error notification instead of failing when the data is too large to turn into one JSON string (still around 400 MB of trace).
• revoke the download Blob URL once the download has started.

Not changed: import still reads the whole file as one string, and traces over roughly 400 MB would need a different file format.

Tests: added round-trip tests for the new format, the legacy list and Uint8List inputs, an in-memory ByteData, a sub-view of a larger buffer, and a size check. Also updated the existing toJson assertion. I have not yet measured the memory difference in a browser against a real 2 minute recording.

Pre-launch Checklist

General checklist

• I read the Contributor Guide and followed the process outlined there for submitting PRs.
• I read the Tree Hygiene wiki page, which explains my responsibilities.
• I read the Flutter Style Guide recently, and have followed its advice.
• I signed the CLA.
• I updated/added relevant documentation (doc comments with ///).

Issues checklist

• I listed at least one issue that this PR fixes in the description above.

Tests checklist

• I added new tests to check the change I am making...

Feature-change checklist

• This PR does change the DevTools UI or behavior and...
  • I added an entry to packages/devtools_app/release_notes/NEXT_RELEASE_NOTES.md.
  • I ran the DevTools app locally to manually verify my changes.

[gemini-code-assist]

Code Review

This pull request optimizes memory usage and file size when saving and loading performance trace data. It introduces base64 encoding for the Perfetto trace binary instead of the legacy JSON array of numbers, while maintaining backwards compatibility. It also optimizes buffer merging in Uint8ListRingBuffer, ensures only the active view of ByteData is encoded, handles potential RangeError during large exports, and defers revoking object URLs on the web to prevent download cancellations. There are no review comments, so I have no feedback to provide.