Fix: Multiple bug fixes

[JanCSEM]

This PR fixes multiple bugs:

1. Scaling factors were previously converted to floating point scalar regardless of the shape, which made the exportBrevitas function error.

This adds a basic check on the shape of the scaling factor tensor, and only converts to a scalar if the tensor has only 1 element.
Added single Conv layer and simple CNN with channel-wise weights quantization models to the tests to validate.

2. In QuantDivider, each node argument is assumed to be a Tensor. For residuals, and multi-branch concatenations, the argument might be a Tuple or a List.

Added support for 1-level nested node arguments.

3. UnrolledMHA was previously returning a single tensor (the attention outputs), which is not consistent with MHA as implemented in pytorch and Brevitas where a tuple is returned (attention outputs, attention weights).

Fixed the implementation to always return a tuple, with the value of the attention weights or None, depending on the provided option.

[JanCSEM]

@Victor-Jung tagging you for visibility