While it would not be too diffiicult to combine the bits with numeric, it would need some extra code (e.g. similar to the old bit twiddling challenge) ;).
Unless performance is critical, It might be easier to convert to a boolean array, concatenate the subsets, and then convert back to a shorter U8 array.
As a first step you need to tell us how your "bits" are represented in your code. Whatever you showed us made no sense:
You get a string. What is the format of the string? Is this a binary string (each byte representing 8 bits) or a string containing characters 0 and 1? You convert it to a U8 array, then into a 2D U8 array with one row, then back to the 1D array you originally had. This makes absolutely no sense.