Scale transform does not fail on non-integers

[vincentqb]

Since Scale assumes integer-valued float-typed signal, giving a float-valued float-typed signal results is not what one would expect for a float-valued float-typed signal. We could either

• update Scale to take float-valued float-typed,
• give a warning if float-valued are detected happens,
• we could rename the transform to IntegerScale, and introduce another FloatScale for this, with a deprecation warning.

The documentation does highlight this property, so we could also leave it at that. :)

[jamarshon]

Removed Scale in #152