Meta Issue: problems with the F# version of F* (currently lagging behind the OCaml version)

[catalin-hritcu]

Turning this into a meta-issue. All these problems with the F# version don't appear in OCaml version:

• [x] Parsing: one additional shift-reduce conflict (#672)
• [x] Parsing: bizare context sensitive parsing for requires True (#252) and dependent tuples (#404)
• [x] Parsing: <> are not parsed as type application in the first definition of a file with the menhir generated fsyacc parser (reported in more detail in #672)
• [x] Hints: not currently supported, requires a JSON parser (#633) -- this prevents verifying the test suite since --use_hints is passed by default by our makefiles. Fixed by @wintersteiger in #840
• [x] Integers: normalization of integer operations requires moving to big ints in .NET, which we should anyway do for soundness (#1054)
• [ ] Floating point literals are badly handled by the F# extraction mechanism (#793)
• [x] Interactive mode not working with the F# version version-to-list: Invalid version syntax: ‘’ (must start with a number)
• [x] Autodeps: F# and OCaml versions used to process files in different order (see #871)
• [ ] No support for --indent (#874)
• [x] FStar.UInt discrepancy (#872)
• [ ] Cannot make ulib (#1023)
• [x] Problem with trailing spaces on set-option (#1051)

[kyoDralliam]

The additional shift-reduce conflict is a known bug in fsyacc (https://github.com/fsprojects/FsLexYacc/issues/39) due to a bad handling of non-associative-ness. The next 2 bugs were caused by bad range queries in the parser and are fixed in kyod_parser.

[cpitclaudel]

I ticked off the "interactive mode" box; I fixed that some time ago.