Add `hypot` specification for computing the square root of the sum of squares

[kgryte]

This PR

• resolves https://github.com/data-apis/array-api/issues/544 by adding a specification for computing the square root of the sum of squares.
• follows IEEE 754 guidance that, if one of the operands is +-infinity, the result is infinity, regardless of whether the other operand is NaN. Intentionally, this deviates from a naive sqrt(x**2 + y**2) implementation.
• only allows underflow when both arguments are subnormal and the correct result is also subnormal. This follows C99 and the POSIX standard and effectively prohibits a naive implementation. The entire purpose of this API is to avoid underflow/overflow during intermediate steps of computation and allowing a naive implementation defeats that purpose.
• follows sqrt in requiring floating-point data types. The guidance uses "should", not "must", to allow implementations to support non-floating-point data types. The reason for the floating-point restriction is that the output data type must be a real-valued floating-point data type and promoting from, e.g., an integral to a floating-point data type is implementation-defined. For defined promotion semantics, a consumer should provide input arrays having floating-point data types.

[leofang]

I am a bit concerned about mentioning subnormal here. In CuPy we always set the compiler flag -ftz=true which would flush all denormals to zero. This helps significantly with floating point operations on GPU. Would this PR imply that we can no longer do this?

[leofang]

Q: I forgot what conclusion we had on this API, @kgryte do we plan to target v2023?

[kgryte]

@leofang I've added a note concerning subnormal behavior and the fact that actual hardware support may vary. Does this allay your concerns?

[rgommers]

Great, looks like everyone is happy now - let's hit the green button. Thanks @kgryte & reviewers!