to_string<float> uses double math and fails with -Werror=float-conversion

[puzrin]

With ETL 20.48.1 and GCC 15.2 / ESP-IDF 6.1, etl::to_string<float>(3.14f, output) fails with:

conversion from 'double' to 'float' may change value [-Werror=float-conversion]

In private/to_string_helper.h (lines 350 and 407), ::floor(...) returns double, which is assigned back to float.

Minimal reproducer and run instructions: https://github.com/puzrin/etl-float-repro

Similar float-formatting improvements appear to have been made in #1379 for format.h, but this path in to_string still uses double math.

[jwellbelove]

Can you try this patch?

etl-1593-to_string-float-uses-double-math .patch

[jwellbelove]

I've checked the changes with your minimal reproducer, and the warnings are gone.

[puzrin]

Thank you. With this patch, the build is ok in my project without kludges.