float32, float128

[yegor256]

Let's create new objects float32 and float128, which should be identical to float, but with smaller/bigger bitwise layout: 32 and 128 bits. Maybe, first we can create them in Java and then change the implementation to EO.