update floating point explaination

[wheresaddie]

needs to cover/rework:

• focus on floating point number issues.
• explain general/transferable reasons instead of suggesting 'permill'
• define rational/real numbers when referring to floats

Originally posted by @wheresaddie in #85