Fail p03 when no bounds checking is used

[mbernat]

When running p03 on macos the check passes even with an incorrect kernel.

    # FILL ME IN (roughly 2 lines)
    output[i] = a[i] + 10

(mojo-gpu-puzzles) chill@Mareks-Laptop mojo-gpu-puzzles % pixi run p03
✨ Pixi task (p03 in default): mojo problems/p03/p03.mojo                                                                                                  
out: HostBuffer([10.0, 11.0, 12.0, 13.0])
expected: HostBuffer([10.0, 11.0, 12.0, 13.0])

This PR catches the mistake by using a bigger array.

(mojo-gpu-puzzles) chill@Mareks-Laptop mojo-gpu-puzzles % pixi run p03              
✨ Pixi task (p03 in default): mojo problems/p03/p03.mojo                                                                                                  
out: HostBuffer([10.0, 11.0, 12.0, 13.0, 10.0, 10.0, 10.0, 10.0])
expected: HostBuffer([10.0, 11.0, 12.0, 13.0, 0.0, 0.0, 0.0, 0.0])
stack trace was not collected. Enable stack trace collection with environment variable `MOJO_ENABLE_STACK_TRACE_ON_ERROR`
Unhandled exception caught during execution: At /Users/chill/Documents/GitHub/mojo-gpu-puzzles/problems/p03/p03.mojo:55:29: AssertionError: `left == right` comparison failed:
   left: 10.0
  right: 0.0

• The incorrect kernel might have a UB, so this is just a best effort check to protect against that.
• This might need additional discussion in the puzzle text itself to make the distinction between thread count and array size clearer.
• Another approach would be to tell the user they likely forgot to check the bounds but hide the bigger array from them.
• Once some form of this PR is accepted I'd be happy to apply it also to other puzzles that suffer from the same problem.

[ehsanmok]

That's by design! Please take a look at the solution. The UB is called out and discussed later.