parse_html.activity: about 30% speedup for html parsing

[karlicoss]

I've been reviewing some older takeouts, so had cachew off and parsing was a bit painful... so had a quick look in a profiler and did some optimization

A few optimizations

• using .find instead of .select is faster since it's not using CSS selectors
• using SoupStrainer is faster since it only does partial parsing and avoids materializing parts of soup we don't actually use

Measurements on a big Chrome/MyActivity.html file

• before

  • parsing (up to for loop over outer_divs: 17s
  • processing (everything in for loop): 16s

• after

  • parsing: 13s
  • procesing: 11s

[purarue]

thanks!

will parse against my old takeouts just to make sure nothing breaks and merge in a bit

[karlicoss]

I was checking against quite old ones, so hopefully should be ok, but thanks for double checking :)