absolute max short bug

[dpaiton]

We have a few functions associated with calculating the absolute max short:

calculate_max_short_upper_bound --> max only considering yieldspace curve
absolute_max_short_guess --> one-shot approximation to the Hyperdrive pool's max short (incl fees & solvency reqs)
calculate_absolute_max_short --> iterative method for refining pool's absolute max short guess

They all seem to have some problems.

• calculate_max_short_upper_bound fails this test:

- tolerance = some small value (ideally zero)
- initialize & trade to a random state on-chain
- ys_max = state.calculate_max_short_upper_bound()
- celine.open_short(ys_max)
- new_max = state.calculate_max_short_upper_bound()
- assert!(new_max <= tolerance)

I'm seeing failures with tolerances on the order of 1e36.

• absolute_max_short_guess does not always a return solvent result

• calculate_absolute_max_short often throws errors from insolvent guesses, resulting in many tests having to catch panics and errors whenever using the function.