int values should be stored as int64 instead of double/float64

**Steps to Reproduce**
Store an int above 2^53 into a Hive box. After reading the value again, it won't be the same, it will have lost some precision since Hive currently converts the value into a double for storage. On JavaScript this still gives correct behavior, since int on that platform is just a double. But on other platforms, it could be improved so that 64-bit integers could have their values preserved without custom adapter work.

**Code sample**
```
test('int', () {
final value = 0xFECDBA0987654321;
var byteData = ByteData(8)
..setFloat64(0, value.toDouble(), Endian.little);
var br = fromByteData(byteData);
expect(br.read(FrameValueType.intT), value);

byteData = ByteData(9)
..setUint8(0, FrameValueType.intT)
..setFloat64(1, value.toDouble(), Endian.little);
br = fromByteData(byteData);
expect(br.read(), value);
});

Expected: <-86208267761138911>
Actual: <-86208267761138912>
```

**Version**
- Platform: iOS
- Flutter version: 2.9.0-1.0.pre.15
- Hive version: 2.0.5

Start with the provided int test and trace from ByteData and fromByteData through FrameValueType.intT. Reproduce the precision loss for the shown value, then verify that storing and reading it preserves the exact 64-bit integer on non-JavaScript platforms.

Written by the indexing model from the issue text.