Add word boundary check to period tokenizer regex.

When only a millisecond based period is provided (e.g. "45 milliseconds"), the period is recognized as minutes ("45m").
This commit adds a word boundary check to the tokenizer regex to distinguish between minutes and milliseconds.

timed.js

@@ -42,7 +42,7 @@
 
 		if (isNaN(parseInt(period,10))) period = '1'+period;
 
-		var match = period.toLowerCase().replace(/[^a-z0-9\.]/g, "").match(/(?:(\d+(?:\.\d+)?)(?:days?|d))?(?:(\d+(?:\.\d+)?)(?:hours?|hrs?|h))?(?:(\d+(?:\.\d+)?)(?:minutes?|mins?|m))?(?:(\d+(?:\.\d+)?)(?:seconds?|secs?|s))?(?:(\d+(?:\.\d+)?)(?:milliseconds?|ms))?/);
+		var match = period.toLowerCase().replace(/[^a-z0-9\.]/g, "").match(/(?:(\d+(?:\.\d+)?)(?:days?|d))?(?:(\d+(?:\.\d+)?)(?:hours?|hrs?|h))?(?:(\d+(?:\.\d+)?)(?:minutes?|mins?|m\b))?(?:(\d+(?:\.\d+)?)(?:seconds?|secs?|s))?(?:(\d+(?:\.\d+)?)(?:milliseconds?|ms))?/);
 
 		if (match[0]) {
 			parsed.when =