Update or remove outdated benchmark instructions from README.rst.

[jdmoorman]

The Benchmarks section of README.rst suggests using this command to run the benchmarks manually (i.e. without tox):

$ python -m pyperf jsonschema/benchmarks/test_suite.py --hist --output results.json

However, this command fails with version 2.0.0 of pyperf:

$ python -m pyperf jsonschema/benchmarks/test_suite.py --hist --output results.json
usage: -m pyperf [-h] {show,hist,compare_to,stats,metadata,check,collect_metadata,timeit,system,convert,dump,slowest,command} ...
-m pyperf: error: argument action: invalid choice: 'jsonschema/benchmarks/test_suite.py' (choose from 'show', 'hist', 'compare_to', 'stats', 'metadata', 'check', 'collect_metadata', 'timeit', 'system', 'convert', 'dump', 'slowest', 'command')

Additionally, there is no such file jsonschema/benchmarks/test_suite.py.

Having a look at tox.ini, I see that the commands used to run pyperf are actually

    perf: mkdir {envtmpdir}/benchmarks/
    perf: {envpython} {toxinidir}/jsonschema/benchmarks/issue232.py --inherit-environ JSON_SCHEMA_TEST_SUITE --output {envtmpdir}/benchmarks/issue232.json
    perf: {envpython} {toxinidir}/jsonschema/benchmarks/json_schema_test_suite.py --inherit-environ JSON_SCHEMA_TEST_SUITE --output {envtmpdir}/benchmarks/json_schema_test_suite.json

which suggests the correct commands for running the benchmarks manually would be

$ python jsonschema/benchmarks/issue232.py --inherit-environ JSON_SCHEMA_TEST_SUITE --output issue232_results.json
$ python jsonschema/benchmarks/json_schema_test_suite.py --inherit-environ JSON_SCHEMA_TEST_SUITE --output json_schema_test_suite_results.json

[Julian]

Fun -- yeah possibly just easiest to only give the option to run them via tox and not try and leak too many implementation details. Happy to see a PR for that if you're up for it.

[jdmoorman]

I've opened #690 which removes the instructions for running tests manually. I'm not sure what to do about the remaining instructions for comparing results between different runs of the benchmarks, since tox -e perf stores those results in {envtmpdir}/benchmarks/issue232.json and {envtmpdir}/benchmarks/json_schema_test_suite.json.

[Julian]

Probably for that piece we just leave it out for now, I meant to go back and implement that possibility but we can just treat that as a feature request same as any other and hit it whenever someone feels it's important enough to make easy. Will have a look at your PR, thanks!