Check if we broke performance by switching from ujson to simplejson

[ara4n]

And whet we'd be better off pinning against a non-broken version of ujson (as apparently the upstream git is fixed).

See #3008

[richvdh]

Empirically, no we haven't broken performance, since matrix.org is now using simplejson without a catastrophic decrease in performance.

I have spent some time profiling ujson, simplejson and json; the results look like this:

(In the above, simplejson uses simplejson.dumps; simplejson_static creates a static JSONEncoder and JSONDecoder and uses encode and decode).

Conclusions are:

• on the load (deserialisation) path, simplejson outperforms ujson for large objects and is comparable for small objects. Stock json is much slower than either.
• on the dump (serialisation) path, the situation isn't as rosy - simplejson is approximately twice as slow as ujson, and 1.5x as slow even as stock json.

Despite that, I think we want to use simplejson anyway. The reasons for this are:

• I think it's important to use a single encoder/decoder consistently, to avoid disasters like the one we had with Replace ujson with simplejson #3008 (where we were able to deserialise an event, but not serialise it again later).
• ujson does not appear to be well-maintained. The bug we found was indeed fixed over a year ago, but that has yet to be released (and I am really not keen to depend on a git version of it, since we'd have to ship our own binaries for debian etc)
• The benchmarks here don't show it, but for some configurations (such as sort_keys=True, as used in canonicaljson), json is way slower than simplejson.
• For me, simplejson's superior load performance wins over json's superior dump performance. We may also be able to push a few optimisations upstream (see Defer is_raw_json test simplejson/simplejson#212, for example).

As part of this work, I've also discovered that canonicaljson is slow (4x slower than simplejson for canonicaljson 1.0.0, though this is reduced to 1.5x in canonicaljson 1.1.0).

[richvdh]

Keeping this open for now, until we merge canonicaljson 1.1.0

[richvdh]

Canonicaljson 1.1.2 is now released, and matrix.org is using it.