Feature Request: Add decimals parameter to specify precision in array_float function

I propose adding a new optional parameter to the array_float function, to allow specifying the number of decimal places for rounding.

This feature would be particularly useful for vector testing, where comparing arrays of floating-point numbers to a specific precision is a common requirement.

Locate the array_float function and inspect its existing callers and tests. Add an optional precision parameter for decimal-place rounding, then update or add coverage that verifies array comparisons at the requested precision.

Written by the indexing model from the issue text.