Correct sd extract_from_text docket number extraction

[grossir]

After checking the parent issue, I realized that the docket numbers in the first page are actually a mix of
"{docket_number}-{disposition}-{author}"

'#30485-aff in pt & rev in pt-JMK\n2024 S.D. 57\n\n    '
'#30590, #30604-dismiss-MES\n2024 S.D. 55\n\n\n\n       '
'#30348-aff in pt & rev in pt-SPM\n2024'
''#30161-aff in pt & rev in pt-SRJ\n2024 S.D. 39\n\n               '

Which matches the current judges

    initial_to_judges = {
        "MES": "Mark E. Salter",
        "SPM": "Scott P. Myren",
        "SRJ": "Steven R. Jensen",
        "PJD": "Patricia J. DeVaney",
        "JMK": "Janine M. Kern",
    }

So, extract_from_text must be corrected, before we re-run it for #1291