QVariant converts floats to string with too much decimal digits

The QVariant in which a float is stored is able to produce its string representation with too much decimal digits with the method toString().

Consider following float: -10795.0684
The following string is generated: -10795.068359375
While this is in fact correct, it's unnecessary information and it causes e.g. Validator on float expecting just 7 decimal digits to fail.

The bug is on the qvariant.cpp - line 433 which sends the float into the QString conversion method: *str = QString::number(d->data.f, 'g', QLocale::FloatingPointShortest);
With double decimal digits requested. The float is converted here to double and hence it gains additional precision!

Instead std::numeric_limits<float>::digits should be used there.