Added translation for LLVM FNeg instruction.

[PhilipBotha]

Added translation for LLVM FNeg instruction. No hint.
Implementation is a modification of the patch by @phoon0416. Using multiplcation by -1.0 instead of subtracting from -0.0.
Added two tests, one for float and one for double that should successfully be analysed with ikos with the FNeg addition without throwing an unsupported LLVM instruction exception.

From the LLVM Language reference manual

The value produced is a copy of the operand with its sign bit flipped. The value is otherwise completely identical; in particular, if the input is a NaN, then the quiet/signaling bit and payload are perfectly preserved.

Using compiler explorer to test:
Using (-0.0) - NaN does not flip the bit. Same with -1.0 * NaN. -NaN does flip the bit.

The code in this PR does not quite adhere to the LLVM requirements of FNeg, but should cover most use cases.

[PhilipBotha]

Outstanding with this PR is the location and running of the two test cases for verifying that the ikos can now handle LLVM FNeg instructions.