[FIRRTL] Cannot index into giant vector (2^32 elements)

("Should indexing operations have 64bit operands?")

We support large vectors, and the FieldIDTypeInterface uses 64bit fields (especially important for vector-of-vector-...).

However indexing operations are only 32bit so we cannot index into these, consider:

```firrtl
circuit Field64:
module Field64:
input x : UInt<1>[4294967296]
output y : UInt
y <= x[4294967295]
```

Which we reject during parsing:
```
field64.fir:5:12: error: value is too big to handle
y <= x[4294967295]
^
```

The same input parses (although don't recommend pushing through the pipeline) if not indexed into (drop last two lines).

Start by reproducing the FIRRTL example with the 2^32-element vector and indexing expression, then trace the parser and indexing handling for the reported 32-bit limit. Done means the example is accepted with a 64-bit index operand while the existing non-indexed large-vector input continues to parse correctly.

Written by the indexing model from the issue text.