Respect byte-order mark.

[nojaf]

This worked on my machine on Windows.
I'm not really an expert when it comes to these things, all suggestions to improve are welcome.

[Smaug123]

I'm not 100% certain that Fantomas should respect BOM, honestly. I lean towards "yes, it should respect BOM", because it seems nice to write out in the same encoding that we read in; but I could understand if Fantomas were opinionated about output format.

[nojaf]

I think it is a good thing to respect the BOM, the case about formatting outside the IDE seems like a valid one to have this.

FSharpSpec says:

All input files are currently assumed to be encoded as UTF-8.

See https://fsharp.org/specs/language-spec/4.1/FSharpSpec-4.1-latest.pdf#page=25
So I'm not sure we should support UTF16 and UT32 as well.
Any thoughts on this @jindraivanek?

@auduchinok how does Rider handle this? Can it work with UTF16 F# files for example?

[Smaug123]

Huh - if the F# compiler is not guaranteed to handle non-UTF8 files then that's a very strong argument that we should at least normalise to UTF-8. Whether to support other encodings as input really boils down to "how hard will it be"; if it's very easy, then it would kind of be nice to receive an arbitrary .txt file with F# code in, and be able to format it. But that does seem low priority if it's easier to assume UTF-8.