[BugFix] Fix LazyStackedCompositeSpec and introducing consolidate_spec

[matteobettini]

This PR will be a superset of #1365, which should be closed.
It tackles tasks from #969
Together with #1391 it becomes a superset of #1204 which should be eventually closed

It fixes #1343

Methods missing are due to time constraint and we can implement them in the future on a need basis

Fixes operations in lazy stack composite specs:

• unsqueeze
• squeeze
• print (it behaves like LazyStackedTd)
• __len__
• __eq__
• __ne__
• project
• type_check
• __delitem__
• __iter__
• keys
• __setitem__
• update
• is_in
• unbind

Missing:

• encode

Bonus, some operations in lazy stacked specs:

• shape
• stack only if same spec type, dtype and ndims
• print
• __len__
• __eq__
• squeeze
• unsqueeze
• unbind

Missing:

• is_in
• project
• type_check
• encode

[matteobettini]

@vmoens for print of het dicts what about something like

LazyStackedCompositeSpec(
    camera: BoundedTensorSpec(
        shape=torch.Size([3, 32, 32, 3]),
        space=ContinuousBox(
            minimum=Tensor(shape=torch.Size([3, 32, 32, 3]), device=cpu, dtype=torch.float32, contiguous=True), 
            maximum=Tensor(shape=torch.Size([3, 32, 32, 3]), device=cpu, dtype=torch.float32, contiguous=True)),
        device=cpu,
        dtype=torch.float32,
        domain=continuous),
    vector: LazyStackedTensorSpec(
         shape=torch.Size([3, -1]), space=None, device=cpu, dtype=torch.float32, domain=continuous), device=cpu, shape=torch.Size([3]),
    
   lazy_keys={
        0 -> lidar: UnboundedContinuousTensorSpec(
                  shape=torch.Size([20]),
                  device=cpu,
                  dtype=torch.float32,
                  domain=continuous),
              obs_0: UnboundedContinuousTensorSpec(
                  shape=torch.Size([1]),
                  device=cpu,
                  dtype=torch.float32,
                  domain=continuous),
        1 -> lidar: UnboundedContinuousTensorSpec(
                  shape=torch.Size([20]),
                  device=cpu,
                  dtype=torch.float32,
                  domain=continuous),
              obs_1: UnboundedContinuousTensorSpec(
                  shape=torch.Size([ 1,2]),
                  device=cpu,
                  dtype=torch.float32,
                  domain=continuous),
        2 ->  obs_2: UnboundedContinuousTensorSpec(
                  shape=torch.Size([ 1,2,3]),
                  device=cpu,
                  dtype=torch.float32,
                  domain=continuous),
    }

 device=cpu, shape=torch.Size([3]))

[vmoens]

Great work!
Aside from the few comments I left, I noticed that the new classes are not part of the doc.

[vmoens]

LGTM
Do the boxes that are unticked raising errors when called?

[vmoens]

LGTM
Do the boxes that are unticked raising errors when called?

[matteobettini]

LGTM Do the boxes that are unticked raising errors when called?

i ll find out and fix if needed

[vmoens]

LGTM Thanks!