CE: Cleanups for conversion warnings

1) If converting from float to int, adds an explicit cast as well as
considers rounding.
2) Adjust some arguments from int32_t to uint32_t where that makes
sense.
3) Adds explicit casts from int32_t to uint32_t where a cast is
necessary.
4) Converts constants used in the tests to the right type (float instead
of double, unsigned instead of signed).

Test: Builds with "-Wconversion" enabled locally
Bug: 129481165
Change-Id: I405af72c22f37865537bdefa3f589b4b2930f56c