Update guess_format_id with new formats

[amoeba]

We're getting some new formats in dataone, https://gist.github.com/amoeba/d4771fc01d4f8f66c44202856d078e8e

• Update guess_format_id's extension<->formatId map
• Make sure ipynb -> json is in there as it's non-obvious
• Can we add MATLAB version detection to the algo? I already did this for netcdf
• What can we do about RAW files? Are there a few common RAW extensions?

[maier-m]

May be difficult to find MATLAB version.
Could potentially use foo <- R.matlab::readMat("file_path")
and attr(foo, "header")$version, but on the few tests I've run all return "5" which is not? a MATLAB file type version. Also v7.3 won't open with the above R package (could use a try call and possibly the h5 package to confirm v7.3 files). Other approaches may be more efficient.

[maier-m]

https://www.loc.gov/preservation/digital/formats/fdd/fdd000241.shtml here would be a good place to look for RAW file extensions. Also potentially just .raw as seen here https://arcticdata.io/catalog/#view/doi:10.18739/A2SC4M