Implement hyperbolic ufuncs

[juntyr]

• Implements the sinh, cosh, tanh, arcsinh, arccosh, and arctanh ufuncs
• Adds tests for the hyperbolic ufuncs
• Adds extra tests for handling -0.0
• Guarantee zero sign sensitive minimum, maximum, fmin, fmax

[juntyr]

@SwayamInSync Once we've figured out a way to handle the trigonometric ufuncs correctly, I'd like to add the hyperbolic ones as well - but I'll wait with writing tests until we have the trigonometric ones figured out

[SwayamInSync]

Got it, thanks @juntyr

[juntyr]

@SwayamInSync Since the hyperbolic ufuncs don't have the same issues as the trigonometric ones, I wanted to get this PR over the line first. I've also added more tests for -0.0 inputs and fixed min/max for them.

[SwayamInSync]

Hmm... hyperbolic were also defined as summation variants of exponentials, did you confirm that they don't have the same problem as #126 on very large inputs say 1e100 or something?

[juntyr]

Hmm... hyperbolic were also defined as summation variants of exponentials, did you confirm that they don't have the same problem as #126 on very large inputs say 1e100 or something?

Let's try that!

[juntyr]

@SwayamInSync So sinh, cosh, tanh, and arctanh handle the large inputs as our tests expect. But for arcsinh and arccosh the floating result can overflow while the quad result doesn't yet. How would you suggest that we check that?

[SwayamInSync]

@SwayamInSync So sinh, cosh, tanh, and arctanh handle the large inputs as our tests expect. But for arcsinh and arccosh the floating result can overflow while the quad result doesn't yet. How would you suggest that we check that?

1e1000 and + are beyound float limits, we are good if they are working till one big common range i.e. 1e100 you can only test till this and we can take this in

[juntyr]

@SwayamInSync From my end this should be ready to be merged now :)

[SwayamInSync]

Merging it, thanks @juntyr