Support for all numpy dtypes

Numpy supports the following dtypes as per [the docs](https://numpy.org/doc/stable/reference/arrays.dtypes.html#specifying-and-constructing-data-types). However `npyjs` only supports the following:

| character | description | supported |
| --- | --- | --- |
| '?' | boolean | NO |
| 'b' | (signed) byte | NO |
| 'B' | unsigned byte | NO |
| 'i' | (signed) integer | YES |
| 'u' | unsigned integer | YES |
| 'f' | floating-point | YES |
| 'c' | complex-floating point | NO |
| 'm' | timedelta | NO |
| 'M' | datetime | NO |
| 'O' | (Python) objects | N/A |
| 'S', 'a' | zero-terminated bytes (not recommended) | NO |
| 'U' | Unicode string | NO |
| 'V' | raw data ([void](https://numpy.org/doc/stable/reference/arrays.scalars.html#numpy.void)) | NO |

Would be great to add these in.