FloatingPointMultiplier to narrower FP type

### Motivation

Currently, the fp multiplier can expand to a wider FP type in both exponent and mantissa.
It throws if we try to have it output a narrower type.

### Desired solution

Allow for conversion to both smaller exponents and/or mantissas by
1) Adding a rounding capability, which can result in a bump in the exponent (and a check for overflow)
2) computing to a narrower exponent may require checking for overflow
3) potentially right-shifting the mantissa in the underflow case

### Alternatives considered

_No response_

### Additional details

_No response_