Enable prange over nodes for Numba backend

[maedoc]

Describe the new feature or enhancement

Provide parallelization at the node level in simulations.

Describe your proposed implementation

The Numba backend has a for loop over nodes in the network here:

    for i in range(weights.shape[0]):

Enabling the backend to switch to numba.prange would enable parallelization over nodes, and the switch would be an option passed in by the user/client of the backend class when instantiating it.

Describe possible alternatives

• Numba supports some automatic parallelization with the gufunc mechanism, but it would require a different template structure from the current backends
• Numba cuda backend could do this as well, but not yet mature enough to be used

Additional comments

This was asked about in the TVB list this morning, https://groups.google.com/g/tvb-users/c/GovsAb-xc1k/m/vCTOQcd_FAAJ

[Bikisome]

hello @maedoc sir i made a pull request on this issue so please review it sir