feat: add stdlib smt collection

[vlopes11]

This commit introduces a collections module for the stdlib. It contains, initially, functions to support Sparse Merkle Tree functionality.

It also adds a mtree_smt_index operation that will fetch a depth/index pair from the advice provider and push onto the operand stack.

[bobbinth]

Thank you! This is a review of the general approach rather than a detailed review.

Overall, I think we can simplify this quite a bit and also make this much more efficient (we should be able to do executed smt::get in less than 100 VM cycles). A few concrete thoughts:

1. Instead of MerkleSmtIndex instruction we should introduce a decorator which would determine all the info we need for the procedure and put this info onto the advice stack. This info could include such things as: depth of the leaf, whether the leaf is a root of an empty subtree, flag values needed for non-deterministic branching, maybe something else. We can name this decorator adv.smtget for now.
2. We could then use adv_push (and/or adv_load) instruction to read the data from the advice stack as needed for by the procedure.
3. We need to make sure that we are handling the case of missing values as I've described a while back. Specifically, if the value is not in the SMT, then we should get back [ZERO; 4] rather than a root of an empty subtree.

There are many ways to structure the procedure itself. Here is one example:

1. execute adv.smtget to determine if the depth of the node and whether the node is a root of an empty subtree.
2. Execute mtree_get to read the node at the specified depth.
3. If we node is a root of an empty subtree, prove that (there could be a couple of ways to do that) and return [ZERO; 4] via the stack.
4. If the node is not a root of an empty subtree, "unhash" it to extract the value portion and to verify that the remaining key is correct. Then return the value via the stack.

In MASM this could look like this (though, I'm probably missing some details):

adv.smtget

adv_push # put the depth onto the operand stack
mtree_get

adv_push # get a flag indicating whether the node is a root of empty subtree
if.true
  # the node is not a root of an empty subtree; verify this and extract the value
else
  # the node is a root of an empty subtree; verify this and return [ZERO; 4]
end

[bobbinth]

Thank you! This is much closer to what we want - but assuming I understood things correctly, it won't quite work just yet. I left some more specific comments/questions inline.

[bobbinth]

Looks good! Thank you! I think the overall structure is good now. I left a few comments inline. I commented only get_16 variant for MASM, but I'm guessing some of these comments could apply to the other variants too.

Finally, we should implement 0xPolygonMiden/crypto#119 first, and then update how advice injector works here.

[bobbinth]

Thank you! I left a couple more test-related comments.

[bobbinth]

Thank you! I left a few more test-related comments inline.

[bobbinth]

Looks great! Thank you! I left a couple of non-blocking comments inline.

[bobbinth]

nit: I think we should add 6 cycles to each value: 2 for adv_push.2 online 286 and 4 for the if/else blocks.

[vlopes11]

Fixed

[bobbinth]

nit: I'd reverse the order of bullet points so that we are describing starting at the top of the stack.

[vlopes11]

Fixed

[bobbinth]

Similar nit as before: I'd reverse the order of bullet points.