Fix wrong Float representation and parsing

[TadeasKucera]

Float values are not parsed / stored correctly.

• Longer floats are trimmed: 0.869932552472 becomes 0.869933, which is undesirable.
• Many zeros behind the dot, like 0.0000000001 are stored as 1e-10 which is also not wanted.

I have added two tests into a new branch fix_float_values which demostrate the problem.

[metthal]

Resolved in #179