128-bit integer encoding and decoding support

[kperryua]

128-bit integer support for new-codable APIs. In the absence of Decimal, which would create a Foundation dependency, some large integers formatted as "floating point" numbers (e.g. contains ., e, or E in JSON) will not parse correctly that do in `JSONDecoder.

Enables tests migrated from Foundation JSONDecoder tests

Addresses #1831

[kperryua]

At some point we may consider an alternative parser that can parse integer-containing floats directly as integers without using Double (or Decimal), but that's far from straightforward.

Decoding integer-containing floats as integers has implications for true stream parsing as well. We might end up being required to buffer input characters for numbers to handle more complex cases. This could incur additional overhead for integer parsing.

Alternatively, we could reject any .eE characters when being asked to decode an integer type. This is what some other systems do, like serde.