Upgrade fast-glob to fix issue on Apple silicon

[bmewburn]

mrmlnc/fast-glob#296

[dotsam]

I've asked the fast-glob maintainer if they could release a new stable version with this fix, and then hopefully you could bump to that version and make a new release? Of course, this is only a workaround and will only use one thread presumably. I guess the real fix will be to get VSCode shipping with a version of node that reports os.cpus() correctly.

For now, I've worked around this by setting intelephense.runtime to a node 15 binary built through NVM, but the homebrew arm node 15 binary should work too most likely.

[shageevan]

Thank you @dotsam. Worked like a charm!

[tvhung83]

Thank you @dotsam , you're a lifesaver, I keep wondering why it works via Rosetta but not natively. I don't want to run the Rosetta version as it costs me so much (precious) RAM.