Use rapidjson for STAC serialisation

[Ariana-B]

Split out from #1065

NaN and Inf values are acceptable in STAC documents but orjson will serialise them to null so switch to using rapidjson instead.

---

📚 Documentation preview 📚: https://datacube-explorer--1095.org.readthedocs.build/en/1095/

[codecov]

Codecov Report

❌ Patch coverage is 83.33333% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 83.58%. Comparing base (ac508cc) to head (01b1863).
⚠️ Report is 1 commits behind head on develop.

Files with missing lines	Patch %	Lines
cubedash/logs.py	66.66%	1 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #1095      +/-   ##
===========================================
- Coverage    83.59%   83.58%   -0.02%     
===========================================
  Files           35       35              
  Lines         4322     4319       -3     
  Branches       546      545       -1     
===========================================
- Hits          3613     3610       -3     
  Misses         507      507              
  Partials       202      202              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[pjonsson]

There are only 2 other calls to orjson, can we replace those as well so we don't have to have two high-performance JSON libraries?

(I think replacing the dumps(..) in logs with dumps(..).encode() and the dumps(..).decode("utf-8") in _filters.py with just dumps(..) should give the same wire-format as we have with orjson.)

[pjonsson]

Can't we tack on an .encode() here and leave __call__ below as it was to get this PR to work?

(I would happily get rid of structlog, but that's for another PR.)

[Ariana-B]

Even when I fix the structlog error by configuring it with WriteLoggerFactory instead of BytesLoggerFactory, I get a TypeError because of click.testing.BytesIOCopy - so structlog isn't necessarily the issue here.

[pjonsson]

Oh, I was digging around in that code for fixing what turned out to be the multiple things closing the same buffer. I know there is one fix on Click in the 8.3.3 milestone that will improve things with testing, but I doubt that fix will help with this.

I don't know how much spare time you have, but if you want to go fix Click, it will probably save you some work if you do that after 8.3.3 is released.