Scanning Architecture Refactor + Additional CacheHandler Documentation

[sonicfind]

• Add DTAEntry type that siphons the old ParseDTA code - providing an avenue to pre-loading information for a DTA node before loading a new song entry.

• Streamline the hierarchy of construction to the most basic form. All the processing of data that produces the member variables for new entries will be performed solely through static functions. Additionally, we will now mark all the values in any of the SongEntry types as readonly. This was already the case for the "ini` subtypes, but the RBCON structures prior to this change required the live modification of an existing entry. The only exception is Song Length, as processing a missing length value requires loading the mixer - which requires setting all subtype information.

• Provide SongMetadata & RBMetadata types with constructors & Serialize() functions holding the code that used to exist in the SongEntry & RBCONEntry constructors and Serialize() functions.

• Streamline the Serialize() functions so that they all share the same signature - allowing for proper virtualization. For the ini subtypes, this meant moving the relative path and EntryType boolean writes to the IniGroup SerializeEntries function instead.

• To add to the above: instead of passing around a BinaryWriter wrapper around the underlying MemoryStream, I've added StreamExtensions to implement directly writing other values to streams - allowing use to pass the streams themselves to all the serialize functions.

• Alter the return types of the Midi & .Chart scanning functions to return the tick resolution of the underlying file. We require that value so that we can calculate the sustain cutoff or hopo thresholds where necessary. By setting the values during the scan, we dodge the necessity of re-evaluating the values at song load time - which also means that the songcache more accurately depicts the true state of the songs it contains.

• Replace the ParseSettings variable in song entries with a trimmed down LoaderSettings variable. This new type only carries the two threshold types and the overdrive midinote. On song load, we can gather all the other values used in ParseSettings through the other states present in the entry instance. Additionally, the appropriate values to use for song loading are calculated during the scanning process instead of before every load. The only adjustments after the fact should be singled out to state changes in the loading code (i.e. possible +1 tick additions).

• Add new ScanResult entries to catch songs that try to be funny and set their resolutions to zero.

• Provide YARGTextContainer with a Null variable similar to FixedArray alongside an IsActive property to allow the simplification of functions that had to use a boolean variable as the return type with a container as an out variable.

• Add DTAEntry type

• Optimize the RBCON update and upgrade utilization:
  -- Firstly, instead of holding all the updates and upgrades in separate lists during the scanning of new items, use the new CONModification type to combine the updates and upgrades that apply to the same DTA node name into a single instance. This , combined with the new DTAEntry type, allows for more-optimal deferment of the processing of DTA data without the downside of re-evaluating the data for song entries that share the same node name. However, upgrades loaded from cache will be held in a separate list.
  -- Secondly, disallow separate SongUpdate folders to hold references to the same node names. Only the update with the most recent write date will be used to modify files. This simplified the DTA parsing scheme so that no looping occurs over multiple update nodes at different locations.

• Re-convert CONFile to a static class that contains a function that generates a List<CONFileListing> from a CON. The "TryGetListing" method becomes an extension on that type.

[EliteAsian123]

HUGE. Unfortunately, I don't really have the time to look through all of the code but a lot of these are good changes overall!