Make `ignore_comments` not default (next major release)

[mverleg]

Stripping comments from files that are multiple megabytes takes an obscene amount of time, even if there are no comments.

I don't think I can make it much faster without writing a whole json parser (in something other than Python, or it can never compete).

So I'm thinking of something like:

• Make ignore_comments=False the default
• Ignore comments only for small files, unless ignore_comments is used to override
• Only parse comments if the file starts with a comment, unless ignore_comments is used to override
• Only strip comments on uncompressed files, unless ignore_comments is used to override

[mverleg]

As a way to address

• Performance
• Excessive deprecation messages

I think it might be good to always try to

• Parse without stripping comments if ignore_comments is None
• Then only if it fails (or ignore_comments is True) re-try with stripping comments

This way

• The deprecation warning is only shown if strip_comments was None and there were actual comments (and only the first time)
• The performance without comments is better, at the cost of performance of implicitly stripping comments

[mverleg]

Released v3.16.1 with this optimization,