Armedia Modifications

• Added an in-memory cache for parent node lookups, so that the importer process doesn't always hit the CMS when looking for them and instead can re-use previous lookups if still available
• Added an in-memory cache for loaded metadata files, so we don't have to load them again if we've done so recently (LRUMap, 100K limit)
• Added support for using cm:name attributes where available when naming objects, as well as when looking for duplicates. This is important to allow export schemes where object names could prove problematic for the host O/S, but not the CMS. Thus, the FS name/path doesn't necessarily reflect the CMS name/path.
• (index) Added support for an XML-based scan index to accelerate scanning duty in the event that remote or slow filesystems are in play. The indexes should contain all the information that would otherwise be generated by the DirectoryAnalyzer
• (index) Added support for in-place import without necessarily mirroring the "balanced-tree" FS structure in the CMS, instead using path information from the indexes to organize the imported items.
• Cleaned up the counter (Bulk status) implementation to avoid double-counts, and to properly separate out files, folders and metadata files during scanning.
• Cleaned up the counters to allow for "freezing" wherein their consumption rate is not calculated over the entire run time, but instead is calculated during the time a specific task that moves that counter runs. Thus, it will report more realistic numbers.