regression: codeblock attributes not preserved by markdown writer

[dsteurer]

The markdown writer in the latest release (pandoc 2.14.0.2) omits in some cases attributes of codeblocks.
Earlier versions (at least pandoc 2.12) preserve the attributes correctly.
(Unfortunately, I don't know which version introduced this regression.)

Example

Input:

~~~~ { .haskell startFrom="100"}
qsort []     = []
qsort (x:xs) = qsort (filter (< x) xs) ++ [x] ++
               qsort (filter (>= x) xs)
~~~~

Output of pandoc -t markdown:

~~~~ haskell
qsort []     = []
qsort (x:xs) = qsort (filter (< x) xs) ++ [x] ++
               qsort (filter (>= x) xs)
~~~~

Expected output: same as input

In this example, the attributes are correctly preserved if another class or an identifier is added to the attributes.

[jgm]

Probably due to #7242

[dsteurer]

Thank you for the quick response!