Allow BigDecimal accept Float without precision

[mrzasa]

Make sure that conversion from a Float works consistently with the conversion from a string representing that float.

Current master state:

pry(main)> BigDecimal(1.2)
ArgumentError: can't omit precision for a Float.
from (pry):22:in `BigDecimal'
pry(main)> BigDecimal('1.2')
=> 0.12e1

After this changes:

pry(main)> BigDecimal(1.2)
=> 0.12e1
pry(main)> BigDecimal('1.2')
=> 0.12e1

When precision is omitted it's set to 0.

Fixes #213
Blocked by #313