fix: payload size with contiguous blocks

[raphaelcoeffic]

When contiguous blocks add up to less than the configured payload size, it would produce blocks larger than the difference between start addresses in 2 payload blocks.

For example (output from readelf -l):

  Type           Offset   VirtAddr   PhysAddr   FileSiz MemSiz  Flg Align
  LOAD           0x030000 0x90000000 0x700002b0 0x115ff0 0x115ff0 R E 0x10000
  LOAD           0x1462a0 0x701162a0 0x701162a0 0x000d8 0x000d8 R E 0x10000
  LOAD           0x155ff0 0x90115ff0 0x70116378 0x00008 0x00008 R   0x10000
  LOAD           0x160000 0x24000000 0x70116380 0x0044c 0x0044c RW  0x10000

In this case, elf2uf2 would produce these blocks (only relevant blocks shown):

Block 4706/4712 @0x70116200 (len=256), Family: ST STM32H7xx
Block 4707/4712 @0x70116300 (len=256), Family: ST STM32H7xx
Block 4708/4712 @0x70116380 (len=256), Family: ST STM32H7xx
Block 4709/4712 @0x70116480 (len=256), Family: ST STM32H7xx
Block 4710/4712 @0x70116580 (len=256), Family: ST STM32H7xx
Block 4711/4712 @0x70116680 (len=256), Family: ST STM32H7xx
Block 4712/4712 @0x70116780 (len=256), Family: ST STM32H7xx

This would cause block 4707 and 4708 to overlap, and thus not be flashed properly.

With this change, the payload is adjusted:

Block 4706/4712 @0x70116200 (len=256), Family: ST STM32H7xx
Block 4707/4712 @0x70116300 (len=128), Family: ST STM32H7xx
Block 4708/4712 @0x70116380 (len=256), Family: ST STM32H7xx
Block 4709/4712 @0x70116480 (len=256), Family: ST STM32H7xx
Block 4710/4712 @0x70116580 (len=256), Family: ST STM32H7xx
Block 4711/4712 @0x70116680 (len=256), Family: ST STM32H7xx
Block 4712/4712 @0x70116780 (len=76), Family: ST STM32H7xx