Compress operator parsing hierarchy

[Gabriella439]

This groups operators that work on different types into a single precedence
level to improve parsing performance. Parsing the precedence of these groups
of operators incorrectly doesn't matter since any expression that mixes them is
ill-typed

This improves performance by about 15% on the benchmark expression from #108

Before:

time                 334.4 ms   (308.6 ms .. 375.3 ms)
                     0.998 R²   (0.995 R² .. 1.000 R²)
mean                 349.8 ms   (340.4 ms .. 364.3 ms)
std dev              14.35 ms   (5.761 ms .. 19.71 ms)
variance introduced by outliers: 19% (moderately inflated)

After:

time                 289.9 ms   (274.6 ms .. 297.9 ms)
                     0.999 R²   (0.998 R² .. 1.000 R²)
mean                 309.3 ms   (301.0 ms .. 318.3 ms)
std dev              11.08 ms   (8.083 ms .. 14.31 ms)

[Gabriella439]

Note that I also tried using Text.Parser.Expression.buildOperatorExpression as well and that made things slower (even slower than the original code)

[Gabriella439]

This also improves heap residency, which goes down from ~5.5 MB to ~4 MB