[Data logging] Odd type error when prepping dict for data logging separately from call

[microbit-matt-hillsdon]

[microbit-matt-hillsdon]

Expression of type "dict[str, int]" cannot be assigned to declared type "Dict[str, str | int | float] | None"
  Type "dict[str, int]" cannot be assigned to type "Dict[str, str | int | float] | None"
    "dict[str, int]" is incompatible with "Dict[str, str | int | float]"
      TypeVar "_VT@dict" is invariant
        Type "int" cannot be assigned to type "str | int | float"
          "int" is incompatible with "str"
    Type cannot be assigned to type "None"

The problem is that log.add takes a dict with union values but you can't pass a dict with a more specific type as dict is invariant (because it's mutable). Consider e.g. if dict added a string valued entry to the dict previously typed as dict[str, float].

So we need a read-only interface for dict if there is one with the right properties, or failing that relaxing the typing here for practical reasons.

[microbit-matt-hillsdon]

Reduced case:

from typing import Union, Dict
stuff: Dict[str, int] = {
  'temperature': 1,
  'sound': 1,
  'light': 1
}
stuff2:  Dict[str, Union[str, int, float]] = stuff

[microbit-matt-hillsdon]

Now live on https://python.microbit.org in v3.0.5