Fixing Symbolic List assignment

[anutosh491]

Explaining the operations/changes that the ASR symbolic pass introduces

1. The pass is intended to deal with list assignment. Now there can a couple cases possible when we think about it which are

i) l1 = [pi, sin(x)] # listconstant assigned to list var
ii) l1 = l2 # list var assigned to another list var

2. Now as discussed here Seg Fault for case involving symbolic lists through llvm backend #2450 (comment) , case ii is always a deep copy and as observed works without any issues. Though this can surely be improved
   Rather than the simple assignment we could have something like
   i) Empty l1
   ii) Appening elements of l1 with basic_heap
   iii) basic_assign between elements of l1 & l2
   But that can be taken care of later

3. The pass essentially deals with case i
   What happens is the following (Snippet 1 is transformed into Snippet 2)

// Snippet 1
    l1: list[S] = [x]

// Snippet2

    # Introduce a placeholder for list[CPtr] just like we have one for CPtr
    _l1: list[CPtr] = [x]
    l1: list[CPtr] = []

    # l1 = [x]
    i: i32 = 0
    for i in range(len(_l1)):
        tmp: CPtr = basic_new_heap()
        l1.append(tmp)
        basic_assign(l1[i], _l1[i])

[anutosh491]

I've added few comments for our understanding and I've also added the failing case raised here (#2450 (comment)) as a test.

[certik]

I thin this is good. Thanks for fixing this! I think you can merge it.

[anutosh491]

Thanks for the review. I think we can go forward with this. If we are looking to customize the symbolic pass even more (if we come across any improvement maybe) let's address it through subsequent PRs.

[anutosh491]

cc @certik I am not sure if I have permissions to merge something untill all requirements are met (I can enable auto-merge squash or rabase when all requirements are met) .
Now that the macos build doesn't pass, I am not sure I can get this in.