fix: strip query string before routing so bookmarkable URLs do not 404 - #81

Closes #80

What

• do_GET / do_POST route on urlparse(self.path).path instead of the raw self.path. URLs like /?range=all and /?range=30d&models=... now hit the index handler instead of falling through to 404.
• Adds urlparse import (stdlib).
• Two regression tests in tests/test_dashboard.py cover both /?range=... and /api/data?_=cachebust.

-from http.server import HTTPServer, BaseHTTPRequestHandler
+from http.server import HTTPServer, BaseHTTPRequestHandler
+from urllib.parse import urlparse

     def do_GET(self):
-        if self.path in ("/", "/index.html"):
+        path = urlparse(self.path).path
+        if path in ("/", "/index.html"):
             ...
-        elif self.path == "/api/data":
+        elif path == "/api/data":

Why

#80 has the full motivation. Short version: the dashboard rewrites the URL with ?range= via history.replaceState, so the first range-button click and any subsequent reload would break.

Verification

$ curl -o /dev/null -w "%%{http_code}\n" "http://127.0.0.1:8080/?range=all"
# before: 404
# after:  200

python -m unittest tests.test_dashboard -v — all 25 tests pass, including the two new ones.