Add support for reading float arrays

[Silas217209]

Adds support for decoding floating point arrays as Vec<f32> using codes_get_float_array. This is useful when the values are needed as Vec<f32> downstream since it avoids reading as Vec<f64> and converting manually afterward.

I added 2 benchmarks, which show that it is slightly faster, although the main improvement should be in downstream application code.

double array reading    time:   [1.5537 µs 1.5576 µs 1.5642 µs]  // preexisting

static double array reading
                        time:   [1.5308 µs 1.5359 µs 1.5436 µs]

static float array reading
                        time:   [1.4714 µs 1.4723 µs 1.4734 µs]

I tried to do everything as close to the existing implementations as possible. I am happy to adjust anything if you prefer a different API shape.