bpo-45636: Un-switch BINARY_OP

[brandtbucher]

In addition to cleaning things up considerably, this also appears to have a tiny positive perf impact:

Details

Slower (12):- logging_format: 6.54 us +- 0.10 us -> 6.75 us +- 0.09 us: 1.03x slower- logging_simple: 5.98 us +- 0.08 us -> 6.17 us +- 0.10 us: 1.03x slower- logging_silent: 113 ns +- 3 ns -> 117 ns +- 3 ns: 1.03x slower
- pathlib: 18.7 ms +- 0.2 ms -> 19.1 ms +- 0.3 ms: 1.02x slower
- pickle_list: 4.35 us +- 0.05 us -> 4.45 us +- 0.05 us: 1.02x slower
- unpack_sequence: 46.6 ns +- 1.5 ns -> 47.3 ns +- 2.0 ns: 1.01x slower
- scimark_monte_carlo: 77.9 ms +- 0.8 ms -> 79.0 ms +- 1.5 ms: 1.01x slower
- float: 81.7 ms +- 0.8 ms -> 82.7 ms +- 1.0 ms: 1.01x slower
- scimark_fft: 335 ms +- 4 ms -> 338 ms +- 2 ms: 1.01x slower
- pickle_dict: 27.7 us +- 0.1 us -> 28.0 us +- 0.1 us: 1.01x slower
- sympy_integrate: 22.0 ms +- 0.1 ms -> 22.1 ms +- 0.1 ms: 1.01x slower
- deltablue: 4.74 ms +- 0.05 ms -> 4.76 ms +- 0.04 ms: 1.00x slower

Faster (24):
- pidigits: 195 ms +- 0 ms -> 188 ms +- 0 ms: 1.04x faster
- regex_dna: 218 ms +- 1 ms -> 210 ms +- 1 ms: 1.04x faster
- crypto_pyaes: 88.7 ms +- 1.1 ms -> 85.9 ms +- 0.6 ms: 1.03x faster
- pyflate: 523 ms +- 5 ms -> 512 ms +- 3 ms: 1.02x faster
- regex_v8: 23.4 ms +- 0.4 ms -> 23.0 ms +- 0.1 ms: 1.02x faster
- unpickle: 14.4 us +- 0.5 us -> 14.1 us +- 0.6 us: 1.02x faster
- nbody: 108 ms +- 1 ms -> 106 ms +- 2 ms: 1.02x faster
- fannkuch: 411 ms +- 4 ms -> 405 ms +- 3 ms: 1.02x faster
- chameleon: 7.54 ms +- 0.07 ms -> 7.43 ms +- 0.07 ms: 1.01x faster
- regex_effbot: 3.23 ms +- 0.07 ms -> 3.19 ms +- 0.11 ms: 1.01x faster
- python_startup_no_site: 5.63 ms +- 0.00 ms -> 5.58 ms +- 0.00 ms: 1.01x faster
- python_startup: 7.65 ms +- 0.01 ms -> 7.58 ms +- 0.01 ms: 1.01x faster
- sqlite_synth: 2.77 us +- 0.06 us -> 2.74 us +- 0.05 us: 1.01x faster
- pickle_pure_python: 365 us +- 3 us -> 361 us +- 2 us: 1.01x faster
- scimark_sor: 152 ms +- 3 ms -> 151 ms +- 3 ms: 1.01x faster
- go: 169 ms +- 2 ms -> 168 ms +- 3 ms: 1.01x faster
- regex_compile: 143 ms +- 1 ms -> 142 ms +- 1 ms: 1.01x faster
- unpickle_list: 5.18 us +- 0.04 us -> 5.14 us +- 0.13 us: 1.01x faster
- pickle: 9.94 us +- 0.10 us -> 9.87 us +- 0.06 us: 1.01x faster
- chaos: 77.9 ms +- 0.8 ms -> 77.5 ms +- 0.7 ms: 1.01x faster
- xml_etree_iterparse: 106 ms +- 1 ms -> 105 ms +- 1 ms: 1.00x faster
- dulwich_log: 66.5 ms +- 0.4 ms -> 66.3 ms +- 0.4 ms: 1.00x faster
- mako: 12.0 ms +- 0.1 ms -> 12.0 ms +- 0.1 ms: 1.00x faster
- 2to3: 270 ms +- 1 ms -> 269 ms +- 1 ms: 1.00x faster

Benchmark hidden because not significant (22): django_template, hexiom, json_dumps, json_loads, meteor_contest, nqueens, raytrace, richards, scimark_lu, scimark_sparse_mat_mult, spectral_norm, sqlalchemy_declarative, sqlalchemy_imperative, sympy_expand, sympy_sum, sympy_str, telco, tornado_http, unpickle_pure_python, xml_etree_parse, xml_etree_generate, xml_etree_process

Geometric mean: 1.00x faster

https://bugs.python.org/issue45636

[markshannon]

There is one warning that needs to be taken care of, otherwise LGTM.