Allow TensorBoard to load despite existing MIME type misconfiguration

[joyceerhl]

For #16072

This comprehensively solves bugs like #15636 (comment) without us having to tell users how to delete problematic MIME types from the Windows registry like https://github.com/microsoft/vscode-python/wiki/Troubleshooting-TensorBoard-integration-in-VS-Code#integrated-tensorboard-sessions

In the screenshot below, you can see that even if I've intentionally misconfigured '.js' files to resolve to a MIME type other than 'application/javascript', TensorBoard is still able to load:

• Pull request represents a single change (i.e. not fixing disparate/unrelated things in a single PR).
• Title summarizes what is changing.
• Has a news entry file (remember to thank yourself!).
• Appropriate comments and documentation strings in the code.
• Has sufficient logging.
• Has telemetry for enhancements.
• Unit tests & system/integration tests are added/updated.
• Test plan is updated as appropriate.
• package-lock.json has been regenerated by running npm install (if dependencies have changed).
• The wiki is updated with any design decisions/details.

[rchiodo]

What happens if this is already there? Does mimetypes just ignore the call?

[joyceerhl]

From https://docs.python.org/3/library/mimetypes.html, "When the extension is already known, the new type will replace the old one. When the type is already known the extension will be added to the list of known extensions."

i.e. if .js is mapped to something else, the existing type will be overwritten by what we specify. For context, TensorBoard has done the exact same thing in their CLI launcher codepath for over a year: https://github.com/tensorflow/tensorboard/blob/ea18bb76ebd515f80a7d99656bb08c045638a991/tensorboard/program.py#L392 We don't benefit from that code because we're launching from their Python API entrypoint, namely the launch() function, so this PR just applies the fix they already have to our codepath as well.