Implement `max/min` on `bigint` without setting highest bits

[jaketrookman]

After looking into a request for testing implementation on the series_test.py (#2651 ) it was discovered that there is an issue with the missing implementation of min/max on bigint. There is a current work around, if you are able to directly set the max/min bits, but as the type is variable, you can not get the max/min in general.

Related to Issue #2377.

[stress-tess]

I'd like to discuss this further with u. My initial thought is there's not a good way to do this without max_bits set since there's not a good identity element.

So the normal workflow for calculating the min value is:

smallest_so_far = max(int)  # the identity for the min operation
for a in arr:
    if a < smallest_so_far:
        smallest_so_far = a

The difficulty is there's no biggest or smallest value for the bigint type. For the example above we could set smallest_so_far = arr[0], but in chpl we use reduce scan ops for this, which require an identity. There's probably a way to make this happen but it's not obvious to me the best way to do this