pgm: adapt to recent blkdev change

With fsGetC becoming a "random address" API, it broke pgm. This commit
fixes it. To avoid adding the weight of a blkdev in pgm, I manage the
read offset directly in pgm.

kernel/fs.asm

@@ -452,15 +452,21 @@ fsGetC:
 	xor	a
 	jp	unsetZ		; returns
 .proceed:
+	push	hl
 	call	fsPlaceH
-	jp	fsblkGetC	; returns
+	call	fsblkGetC
+	pop	hl
+	ret
 
 ; Write byte A in handle (IX) and advance the handle's position.
 ; Z is set on success, unset if handle is at the end of the file.
 ; TODO: detect end of block alloc
 fsPutC:
+	push	hl
 	call	fsPlaceH
-	jp	fsblkPutC	; returns
+	call	fsblkPutC
+	pop	hl
+	ret
 
 ; Mount the fs subsystem upon the currently selected blockdev at current offset.
 ; Verify is block is valid and error out if its not, mounting nothing.

kernel/pgm.asm

@@ -57,11 +57,13 @@ pgmRun:
 	push	hl		; unparsed args
 	ld	ix, PGM_HANDLE
 	call	fsOpen
-	ld	hl, PGM_CODEADDR
+	ld	hl, 0		; addr that we read in file handle
+	ld	de, PGM_CODEADDR	; addr in mem we write to
 .loop:
 	call	fsGetC		; we use Z at end of loop
-	ld	(hl), a		; Z preserved
+	ld	(de), a		; Z preserved
 	inc	hl		; Z preserved in 16-bit
+	inc	de		; Z preserved in 16-bit
 	jr	z, .loop
 
 	pop	hl		; recall args