Type system giving error when replacing function's argument type with identical type variable

Loving TypeScript, but also math, thus numpy, thus needing python, I'm currently checking out how much typing can be brought to the python world. It's limited, it seems, and I have a particularly weird case, which I hope somebody can explain to me:

Using a module like

# testModule.py
import numpy.typing as npt

coordsType = list[tuple[float, float]] | list[list[float]] | npt.NDArray

def moduleFuncExplicit(arg: list[tuple[float, float]] | list[list[float]] | npt.NDArray):
    return arg

def moduleFuncWithTypeVar(arg: coordsType):
    return arg

and using these functions from another file like

# testMain.py
import testModule
import numpy as np

arg: list[tuple[float, float]] = [(1.,2.), (3.,4.), (5.,6.)]
npArg = np.array(arg, dtype=np.float64)
npArgAnyDtype = np.array(arg)

ret_Explicit_Simple = testModule.moduleFuncExplicit(arg)
ret_Explicit_Numpy = testModule.moduleFuncExplicit(npArg)
ret_TypeVar_Simple = testModule.moduleFuncWithTypeVar(arg)
ret_TypeVar_Numpy = testModule.moduleFuncWithTypeVar(npArg) # type error
ret_TypeVar_NumpyAny = testModule.moduleFuncWithTypeVar(npArgAnyDtype)

gives me a type error (only) for the fourth line:

Argument of type "NDArray[float64]" cannot be assigned to parameter "arg" of type "coordsType[Unknown]" in function "moduleFuncWithTypeVar"
  Type "NDArray[float64]" cannot be assigned to type "coordsType[Unknown]"
    "NDArray[float64]" is incompatible with "list[tuple[float, float]]"
    "NDArray[float64]" is incompatible with "list[list[float]]"
    TypeVar "_DType_co@ndarray" is covariant
      TypeVar "_DTypeScalar_co@dtype" is covariant
        Type "float64" cannot be assigned to type "ScalarType@NDArray"PylancereportGeneralTypeIssues

1. What does this error mean? Why does it occur?
2. More importantly, why the hell are the two module functions not exactly identical?