Added base parameter to math.log

[Chillee]

Stack from ghstack:

• Added degrees, radians, ldexp #21131 Added degrees, radians, ldexp
• Added frexp, isinf, isnan, isfinite #21130 Added frexp, isinf, isnan, isfinite
• Added factorial #21129 Added factorial
• Finished trigonometric functions #21128 Finished trigonometric functions
• Finished the high-priority functions #21127 Finished the high-priority functions
• Added remainder of high-priority trigonometric math ops #21126 Added remainder of high-priority trigonometric math ops
• Added better tests for math ops and unified them #21125 Added better tests for math ops and unified them
• made floor/ceil return ints #21124 made floor/ceil return ints
• Added base parameter to math.log #21151 Added base parameter to math.log

Differential Revision: D15563185

[Chillee]

Sorry for continually remaking this PR, but I think I messed up something small with ghstack that causes this PR to be constantly remade. Can someone stamp this so I can hopefully stop constantly remaking this PR?

[ailzhang]

Here might fail the lint check

[Chillee]

It passes the precommit lint check - I think it's fine.

That code is also removed in a couple commits later in the stack.

[ailzhang]

LGTM

[xuhdev]

What happens if b == 1? Perhaps we should at least add a test for that case?

[facebook-github-bot]

@Chillee merged this pull request in 8002030.

[Chillee]

@xuhdev I've done that later up in the stack (I unified all the math tests and added much more exhaustive checkign): #21125

But to answer your question, we return inf, which isn't ideal. Unluckily, there doesn't seem to be any easy way to handle these errors, so we've decided to just fail silently (+ inf is a natural value to return, although it doesn't conform exactly with the python API)