feat: Allow markup around inline field names.

This puts the parser on par with the reference one which is [Dataview's inline-field.ts](https://github.com/blacksmithgu/obsidian-dataview/blob/81ba6a0dd31d6562de852144112922bb33e084d9/src/data-import/inline-field.ts#L159).

Support for emojis is left behind.

Fixes selimrbd#16.

pyomd/metadata.py

@@ -421,11 +421,15 @@ class InlineMetadata(Metadata):
             metadata dictionary
     """
 
+    KEY_PART_REGEX = r"([0-9\p{Letter}\w\s_/-]+)"
+    MARKUP_CHARS_REGEX = r'[_*~`]*'
+    FULL_LINE_KEY_REGEX = r"[^0-9\w\p{Letter}]*" + KEY_PART_REGEX + MARKUP_CHARS_REGEX
+
     TMP_REGEX = Template(r"(?P<beg>.*?)(?P<key>$key)::(?P<values>.*)")
     TMP_REGEX_ENCLOSED = Template(
         r"(?P<beg>.*?)(?P<open>[(\[])(?P<key>$key)::(?P<values>.*?)(?P<close>[)\]])(?P<end>.*)"
     )
-    REGEX = re.compile(TMP_REGEX.substitute(key="[A-z][A-z0-9_ -]*"))
+    REGEX = re.compile(TMP_REGEX.substitute(key=FULL_LINE_KEY_REGEX))
     REGEX_ENCLOSED = re.compile(TMP_REGEX_ENCLOSED.substitute(key=".*?"))
 
     def to_string(
@@ -513,7 +517,7 @@ def _parse_1(cls, note_content: str) -> MetaDict:
 
         tmp: dict[str, list[str]] = dict()
         for m in matches:
-            k = m.group("key").strip()
+            k = re.sub(cls.MARKUP_CHARS_REGEX, '', m.group("key")).strip()  # Remove markup characters and strip whitespace
             v = m.group("values")
             tmp[k] = tmp.get(k, "") + ", " + v
         metadata: MetaDict = {