Add float.to-fixed and float.to-precision

[crai0]

Add two new float to string conversion methods that mimic JavaScript's Number.toFixed() and Number.toPrecision().

They are implemented as no_mangle functions similar to the already existing float to shared string conversion.

Closes #5822

[crai0]

I noticed that the mcu_esp tests are not building because f64::abs isn't available in core in version 1.82.0 which the Xtensa toolchain is using. It is part of core since version 1.85.0 though. Is this a reason to raise MSRV or should I rather implement it by setting the sign bit using from_bits/to_bits?

I haven't looked into why the cpp_test_driver fails with a mismatch in two cases yet but I guess it has to do something with floating point precision.

[ogoffart]

Thanks for the PR. I'll review later because I'm busy this week.
But the float issue on embedded is because some functions don't exist on that platform. But we can fix it by adding a use num_traits::Float; like we do in a few places already.