sys.monitoring branches for `match` cases showing incorrectly

Bug report

Bug description:

With the code from PR 122564, the destinations of branches into the blocks of certain cases are mapping to the pattern as well, making them look like they go out of scope:

v = 1
match v:
    case 1:
        x = 1
    case 2:
        x = 2
    case 3:
        x = 3
x += 2

This code yields (branches listing below extracted from co_branches()):

  0          0       RESUME                   0

  1          2       LOAD_CONST               0 (1)
             4       STORE_NAME               0 (v)

  2          6       LOAD_NAME                0 (v)

  3          8       COPY                     1
            10       LOAD_CONST               0 (1)
            12       COMPARE_OP              88 (bool(==))
            16       POP_JUMP_IF_FALSE        5 (to L1)
            20       NOT_TAKEN
            22       POP_TOP

  4         24       LOAD_CONST               0 (1)
            26       STORE_NAME               1 (x)
            28       JUMP_FORWARD            19 (to L3)

  5   L1:   30       COPY                     1
            32       LOAD_CONST               1 (2)
            34       COMPARE_OP              88 (bool(==))
            38       POP_JUMP_IF_FALSE        5 (to L2)
            42       NOT_TAKEN
            44       POP_TOP

  6         46       LOAD_CONST               1 (2)
            48       STORE_NAME               1 (x)
            50       JUMP_FORWARD             8 (to L3)

  7   L2:   52       LOAD_CONST               2 (3)
            54       COMPARE_OP              88 (bool(==))
            58       POP_JUMP_IF_FALSE        3 (to L3)
            62       NOT_TAKEN

  8         64       LOAD_CONST               2 (3)
            66       STORE_NAME               1 (x)

  9   L3:   68       LOAD_NAME                1 (x)
            70       LOAD_CONST               1 (2)
            72       BINARY_OP               13 (+=)
            76       STORE_NAME               1 (x)
            78       RETURN_CONST             3 (None)
'<module>' branches:
    ex8.py 3:9-3:10 "1" (<module>@16) -> 3:9-3:10 "1" (<module>@22) [case -> out]
    ex8.py 3:9-3:10 "1" (<module>@16) -> 5:9-5:10 "2" (<module>@30) [case -> next case]
    ex8.py 5:9-5:10 "2" (<module>@38) -> 5:9-5:10 "2" (<module>@44) [case -> out]
    ex8.py 5:9-5:10 "2" (<module>@38) -> 7:9-7:10 "3" (<module>@52) [case -> next case]
    ex8.py 7:9-7:10 "3" (<module>@58) -> 8:12-8:13 "3" (<module>@64) [case -> body]
    ex8.py 7:9-7:10 "3" (<module>@58) -> 9:0-9:1 "x" (<module>@68) [case -> next]

@markshannon @nedbat

CPython versions tested on:

CPython main branch

Operating systems tested on:

macOS

Linked PRs

• gh-130627

Reproduce the branch listing with the match-case example and inspect the sys.monitoring co_branches() entry point, along with the behavior introduced by PR 122564. Compare the reported destinations for case patterns and bodies; done means branches into case blocks no longer map to the pattern locations or appear to leave scope incorrectly.