2023-08-17: release 1.0.21

* Added string and list functions to the equation language. The new functions
have been added to the FLAGS menu and CATALOG->FCN in the equation editor.
List literals have the format {A:B:C}. List elements may be any type,
including lists or matrices.
* Equations: Fixed stack access within matrix literals.
* Fixed LASTX option in the equation editor's stack menu (R↓ key).
* Fixed Y^X with units where X is a negative reciprocal.
* →LIST and LIST→ didn't print in TRACE mode. Fixed.

2023-08-11: release 1.0.20

* Added 1LINE and NLINE functions, to select 1-line or n-line mode. 1-line mode
forces matrices and lists to be displayed in compact format, even if the
display has more than 2 lines; n-line mode is the multi-line matrix and list
display that was introduced in 1.0.17. The latter remains the default.
* INPUT would crash if R/S was pressed while the stack was empty, in NSTK mode.
Fixed.
* ANUM would consider a decimal point or comma by itself as zero, instead of
ignoring it. Fixed.
* Added UNIT? to the UNIT.FCN menu.
* If an error occurs while evaluating an equation, which does not exist in the
EQN list, going to EQN mode and pressing R/S will now add the equation to the
EQN list, and then start editing with the cursor on the site of the error.
This used to work only for equations that were in the EQN list beforehand.
* Added LASTX for equations, to work alongside REGX/REGY/REGZ/REGT, for
read-only access to the RPN stack.
* EDIT, EDITN, and INDEX didn't clear flags 76 and 77 (matrix edge wrap and end
wrap), while they should. Fixed.
* Decimal version: In HMS+ and HMS-, fixed handling of cases where either or
both of the arguments has a number of minutes or seconds >= 60. (The binary
version did not have this bug.)
* Binary version: max digits for SHOW increased to 17. This was necessary
because 16 digits was actually never enough to begin with. For example, 5 1/X
and 5 SQRT X^2 1/X both returned a result that looked like 0.2 in SHOW, even
though they are different, as can be easily verified with X=Y? or by
subtracting them. It takes 17 decimal digits to ensure that different binary
numbers will always be rendered as different decimal numbers.