-Output is a 12-bit value and should wrap
-"Magnitude"/"step size" should be reset to 0 when playback starts
This fixes sound effects in NEXZR that were broken by the previous (incorrect) fix and also broken because _magnitude was not being reset to 0 when each sample started playing
-Labels defined within the assembler that were referred to before their definition did not work properly and ended up being assigned the wrong value (when the label's address wasn't a value between 0x100 and 0xFFFF) - this is fixed by running multiple passes until the resulting byte code stops changing.
-Fixed issues where e.g EOR $00, Y would be assembled as a 2-byte instruction, which is invalid.
Save ram is not mirrored across the entire 8kb, can only be accessed in the first 2kb
This fixes Death Bringer corrupting save ram when trying to start a new game
When VDC1 had both BG and sprites disabled, it blocked the VDC2's background (resulting in a black screen). VDC1's output should count as transparent in this scenario
+ Automatically move to current address when clicking on the source view tab (so that the active row is shown after manually loading a symbol file, etc.)
The RCR counter doesn't appear to increment unless the HDW mode (e.g the visible scanline) actually starts. On the other hand, the vertical mode counter will get clocked even if the VDC doesn't hit HDW (trigger seems to be earlier, currently set to be the same timing as the Y scroll latch). The RCR IRQ could occur later than on hardware because of this
-IRQ/SCSI status should happen after seek is done when a play command is sent (fixes audio sync in Brandish intro cutscene)
-Fixed seek delay not being applied correctly when swapping from audio playback to data loading and tweaked seek delays to make them a bit slower (fixes Brandish audio glitch during introduction)
-Fixed regression that causes the "transfer ready" irq flag to not be reset when a sector was done reading (this broke the intro sequence in "It came from the desert")
-Improved behavior when a load operation is cancelled (to match the expected results of both scsitest and verificator tests)
-ADPCM DMA speed now matches what krikzz's test rom expects
-Implemented some delays in CD-ROM interface that mostly match the test rom
-Implemented delay when software fails to read an entire sectore before the next one is ready (improves audio sync in Sherlock Holmes movies)
-Re-fixed issues with Ys IV that were apparently caused by the read command not properly updating the SCSI signals after the drive starts seeking to the requested sector
-Fixed some ADPCM playback flags/timing issues based on okitest test rom (playing/busy flag gets turned off slightly delayed vs end reached flag, etc.)
-Implemented delay for $1805 writes, based on test rom
-ADPCM edge cases work better (reset, length latching, dma, etc.)
-BRAM unlocking/locking fixes
-Read ToC command returns the correct amount of data
-SCSI data port returns the drive's reply whenever a reply is being read, regardless of the value written to the port through $1801
-Implemented/fixed audio sample read behavior (latching, L/R toggle)
-Fixed missing (unused) readable/writable bits in some registes
All of these fixes are based on the test results from krikzz's test rom here: https://github.com/krikzz/turbo-ed-pro-pub/tree/main/tcd-verificator
