Use CSS classes for seamless split <code> tags around Fortran %

The previous split <code> approach preserved % but created a visible
gap due to border/padding on each element. Add f90l/f90r CSS classes
that strip internal borders and padding so adjacent code spans render
as one seamless pill: <code class="f90l">bc_y%</code><code class="f90r">beg</code>.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Author: sbryngelson

docs/custom.css

@@ -3,6 +3,22 @@
  * Overrides for doxygen-awesome theme
  */
 
+/* Seamless split <code> tags for Fortran % struct accessors.
+ * Doxygen consumes %<word> even inside code spans, so we split around %
+ * into adjacent <code> elements and remove internal borders/padding. */
+code.f90l {
+    border-right: 0;
+    border-top-right-radius: 0;
+    border-bottom-right-radius: 0;
+    padding-right: 0;
+}
+code.f90r {
+    border-left: 0;
+    border-top-left-radius: 0;
+    border-bottom-left-radius: 0;
+    padding-left: 0;
+}
+
 /* Fix inline code visibility in colored admonition blocks (warning, attention, important, note, etc.) */
 
 /* Warning/Attention/Important blocks (red/pink background) */

toolchain/mfc/gen_physics_docs.py

@@ -88,15 +88,18 @@ def _format_message(msg: str) -> str:
     msg = re.sub(r"for (support|pulse) = \1\b", r"for the given \1", msg)
     msg = re.sub(r"for (support|pulse) = (\d+)", r"for \1 = \2", msg)
 
-    # Wrap code-like tokens in backticks (or <code> for Fortran % accessors).
-    # Doxygen treats %<word> as "suppress auto-link" and consumes the %,
-    # even inside backtick and <code> spans. To preserve the Fortran %
-    # accessor, split into adjacent <code> tags so % is never followed
-    # by a word character: <code>bc_y%</code><code>beg</code>.
+    # Wrap code-like tokens in code formatting.
+    # Doxygen consumes %<word> as "suppress auto-link" even inside backtick
+    # and <code> spans. For Fortran % accessors, use split <code> tags with
+    # CSS classes that remove internal borders, making them visually seamless.
     def _wrap_code(match: re.Match) -> str:
         token = match.group(1)
         if "%" in token:
-            return "<code>" + token.replace("%", "%</code><code>") + "</code>"
+            parts = token.split("%")
+            result = parts[0]
+            for part in parts[1:]:
+                result += '%</code><code class="f90r">' + part
+            return '<code class="f90l">' + result + "</code>"
         return f"`{token}`"
 
     msg = _CODE_RE.sub(_wrap_code, msg)