Closes #3870: bug in reshape for bigint type

[ajpotts]

Fixes a bug where reshape function gives the incorrect result for the bigint type, as shown in the following example:

In [3]: dtype = "bigint"

In [4]: size = 10

In [5]:         x = ak.arange(size, dtype=dtype).reshape((1, size, 1))

In [6]: x
Out[6]: array([array([array([0]) array([1]) array([2]) array([3]) array([4]) array([224251899497]) array([1919968626]) array([78401563153491]) array([127961796865894]) array([111052610564705])])])

Fixes the issue by creating a separate flatten and unflatten function in AryUtil.chpl for the bigint case.

Closes #3870: bug in reshape for bigint type
Closes #3908: flatten for bigint

[ajpotts]

I believe the cause of the bug is this section the unflatten procedure:

            get(
              c_ptrTo(unflat[low]),
              getAddr(a[flatSlice.low]),
              locInStart,
              c_sizeof(t) * flatSlice.size
            );

I think the fact that BigInt does not have constant size, the data cannot be bulk copied using the address.

The fix was write a separate unflatten procedure for the bigint case.

[drculhane]

Looks good. You'll want to rename test_reshape_bug_reproducer :)

[jaketrookman]

Reapproving after formatting changes