MI: fix House votes never scraped due to regex mismatch and tab-separated names (#5696)

Two bugs introduced by #5379 (merged 2025-05-01) that together cause all
Michigan House vote records to be silently dropped.

Bug 1 — regex mismatch (scrape_votes, line 197):
The roll call regex `r"Roll Call # (\d+)"` requires a space between "#"
and the number. House journal action strings use "Roll Call #44" (no
space); Senate uses "ROLL CALL # 1" (with space). Because `rcmatch` is
always None for House actions, the `if rcmatch:` block never executes
and no House votes are ever scraped. Fixed with `r"Roll Call #\s*(\d+)"`.

Bug 2 — tab-separated voter names (parse_roll_call, line 351):
House journal <p> elements separate voter names with single tab characters
("Alexander\tFarhat\tMartus\tSchmaltz"), while Senate journals use
multi-space padding. The split `re.split(r"(?<!,)\s{2,}", p)` requires
2+ consecutive whitespace characters, so single tabs go unsplit — all
four names on a line are stored as one fused string. Note: a
session-specific workaround for the identical tab issue already existed
for 2017-2018 but was never generalised. Fixed by changing the split to
`r"\t|(?<!,)\s{2,}"`, which handles both formats.

Tested against 45 Michigan 2025-2026 bills (38 House, 7 Senate) from a
known-missing dataset. 31 House votes recovered with correct counts and
fully split individual voter name lists (e.g. HB4707: 57 yeas, 44 nays).
Senate bills with no floor votes correctly return 0 votes.

Co-authored-by: Ramon Perez <ramon.perez@digitaldemocracyproject.org>
Co-authored-by: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>

Author: 3 people

scrapers/mi/bills.py

@@ -194,7 +194,8 @@ def scrape_votes(self, bill: Bill, page: lxml.html.HtmlElement):
                 actor = "upper"
 
             # Process roll call votes
-            rcmatch = re.search(r"Roll Call # (\d+)", action, re.IGNORECASE)
+            # House format: "Roll Call #44" (no space); Senate: "ROLL CALL # 1" (space)
+            rcmatch = re.search(r"Roll Call #\s*(\d+)", action, re.IGNORECASE)
             if rcmatch:
                 rc_num = rcmatch.groups()[0]
                 vote_url = None
@@ -347,8 +348,8 @@ def parse_roll_call(self, url, rc_num, session):
             elif p.startswith("In The Chair:"):
                 break
             elif vtype:
-                # Split on multiple spaces not preceded by commas
-                for line in re.split(r"(?<!,)\s{2,}", p):
+                # Split on tabs (House journals) or multiple spaces (Senate journals)
+                for line in re.split(r"\t|(?<!,)\s{2,}", p):
                     if line.strip():
                         if session == "2017-2018":
                             for leg in line.split():