Changes in DOWN 4R5G

 - If ASCOUT set in DOWNDEF (default) and SPEC3 output created, force output
   file/elt to be ASCII if 1) either input is ASCII; 2) either input is an SDF
   file other than standard ('S'), such as a print file; or 3) a TOC/PCF/TCF
   downdate is being performed to a single output (vs. a program file). In a
   program file downdate, the output for each pair of elements will be treated
   as in 1) above.
        If ASCOUT is 0, DOWN will generate output as before, which sometimes
   resulted in an element containing ASCII images (0420000000001) having a
   Fieldata SDF label control word (0500100000000). This could throw off some
   editors and cause them to convert ASCII images to Fieldata.

 =======================================
 Changes in DOWN 4R5F

 - Update GETPFTMSTMP$ routine and BSP$ 77R1 to avoid direct ER TDATE$
   detectable by scanning routines; value is either stored or executed by
   index.
 - When DOWNing TOCs, allow ±hour offset to be applied to first file's elts to
   match those of second file, in case their timestamps show zone differences.
   See DOC, SPEC6, TOC Downdating.

 =======================================
 Changes in DOWN 4R5E*

 - Preset timestamp for SPEC3 elt in case site doesn't have ER TIMEBYINDEX$;
   use GETPFTMSTMP$ routine to retrieve ER PFI$ default timestamp format.
 - Collect release version of DOWN with SYSLIB 77R1 version of BSP$.
 - BSP$ interface level 3 not needed; exclude TIMESUBR$ & entry points from MAP.
 * [Updated to replace GET$$TMSTMP$ routine; other code not changed.]

 =======================================
 Changes in DOWN 4R5D

 - Improve image comparison time by using BICL instruction.
 - Upgrade to STM$PKG Rev. 10 to support full MODSWTIME$ range.
 - Modify SCRSIZ$ so screen read wait activity not overwritten due to timing.

 =======================================
 Changes in DOWN 4R5C

 - DOWN could output garbage characters when switching between Fieldata & ASCII
   on the new 8300/SAIL machines. Some old UOM EDIT$ routines were revised to
   avoid a problem when interpreting the Quarter-Word Mode bit (DB32).

 =======================================
 Changes in DOWN 4R5B

 - Fix bug: DOWN was not correctly counting the number of element pairs with
   differences if the first element in a pair was empty or had only data images
   but no correction cards (-n,m). The value set into T3 of the condition word
   by the Y option was also incorrect. (DOWN output was correct, however.)
 - Clarify Y option in DOC: will not include null entries forced by E option.
 - Update to STM$PKG Rev. 8.

 =======================================
 Changes in DOWN 4R5A

 - Switched to STM$PKG Rev. 7 for all date/time editing and comparisons. Will
   recognize modified SWTIME words if they occur where TDATE$ words are
   expected.

 =======================================
 Changes in DOWN 4R5

 - Will maximize SDFI & index buffers when DOWNing other than large program
   files or T option. DOWN can get I/O errors if there are very large records
   and relatively small buffers. Will also maximize for small program files.
 - Default # SDFI buffers: 4 (single file/elt downdate), 8 (ProgFl-ProgFl).
 - Add H option to show memory acquisition history for SPEC7 parameters.
 - Add ability to display instruction/cycle count used by main processing loop.
 - Add ability to display progress lines (@@X C) to terminal if breakpointed.
 - Was not getting correct highest sector+1 for data file compares. This could
   cause an error or premature end of file.
 - Allow for supplying an initial progress interval in SPEC7.
 - Fixed bugs handling continuation records (051) of long images, particularly
   when they end on or cross over a buffer boundary.
 - Changing @@X C time interval (@@X C,nn) was producing too many messages.