Fix metadata removal on upload for files with un-sane characters in them

When re-uploading a file with characters that will be later sanitized,
the old metadata entry wasn't removed because it checked the unsane
filename rather than the sanitized version that will later be stored.

This was creating an inconsistent metadata view during analysis for
files with spaces in them.

octoprint/gcodefiles.py

@@ -123,14 +123,15 @@ class GcodeManager:
 		if absolutePath is None:
 			return None
 
-		if file.filename in self._metadata.keys():
+		basename = os.path.basename(absolutePath)
+		if basename in self._metadata.keys():
 			# delete existing metadata entry, since the file is going to get overwritten
-			del self._metadata[file.filename]
+			del self._metadata[basename]
 			self._metadataDirty = True
 			self._saveMetadata()
 		file.save(absolutePath)
-		self._metadataAnalyzer.addFileToQueue(os.path.basename(absolutePath))
-		return self._getBasicFilename(absolutePath)
+		self._metadataAnalyzer.addFileToQueue(basename)
+		return basename
 
 	def getFutureFilename(self, file):
 		if not file: