fix[cartesian]: don't upcast arguments of power function

[romanc]

Description

@FlorianDeconinck / @twicki here's the branch/PR which adds an exception for casting exponent and base of the power function to the same type.

Background: This came up in physics validation where we noticed that the current rule set will translate the following stencil code

bquad = (
    2.0
    * rbuoy
    * constants.MAPL_GRAV
    * cridis
    * (thv_x1 - thv_x0)
    / thv0j
    - 2.0 * wue**2
)

into

bquad_gen_0 = ( ( ( ( ( ( np.float32(2.0) * np.float32(1.0) ) * np.float32(9.80665) ) * cridis_gen_0 ) * ( thv_x1_gen_0 - thv_x0_gen_0 ) ) / thv0j_gen_0 ) - ( np.float32(2.0) * ufuncs.power(wue[i + 0,j + 0,k + 0],np.float32(np.int32(2))) ) )

gtOUT__bquad_gen_0 = ((((((float(2.0) * float(1.0)) * float(9.80665)) * gtIN__cridis_gen_0) * (gtIN__thv_x1_gen_0 - gtIN__thv_x0_gen_0)) / gtIN__thv0j_gen_0) - (float(2.0) * dace::math::pow(gtIN__wue, float(int(2)))));

with the debug and dace:cpu_kfirst backends respectively.

The PR attempts to avoid casting the exponent (2) into a 32-bit float just because the base value wue is a FloatField.

Requirements

• All fixes and/or new features come with corresponding tests.
• Important design decisions have been documented in the appropriate ADR inside the docs/development/ADRs/ folder.
  N/A

[FlorianDeconinck]

This is looking good to me. Should we move it in?

[romanc]

This is looking good to me. Should we move it in?

Sure, feel free to go ahead. Then let's see next week if this helps with physics stability on MacOS.