Change way stream blocks are computed to maximize compression.

[pete4abw]

• Add Dictionary Size computation and setting capability.
• Remove 7/9 scaling for lzma levels. Default still 5.
• LZMA levels now 1-9.
• Improve info displays.
• Assembler will now select either nasm or yasm.
• ETA will be displayed after each chunk is processed.
• Decompression status updated every 5 seconds or so.
• Documentation updates.
• Code cleanups (whitespace removal, comment alignment).

[pete4abw]

Some notes about this PR. Please test, review, benchmark, and comment. Because of the nature and depth of changes, recommend bumping version. I set as 0.640Beta.

• File format does not change so this does not impact past version 0.6 lrzip.
• LZMA Compression levels are now 1-9. The 7/9 scaling is removed.
• Stream block allocation logic changed to favor dictionary size changes over block size reduction.
• LZMA Overhead computation changed so that dictionary size can be set for all levels 1-9.
• See ChangeLog and WHATS-NEW for more detail.

  if (LZMA_COMPRESS) {
      if (control->dictSize == 0)
          switch (control->compression_level) {
          case 1:
          case 2:
          case 3:
          case 4:
          case 5: control->dictSize = (1 << (control->compression_level * 2 + 14));
              break; // 65KB to 16MB
          case 6:
          case 7: control->dictSize = (1 << 25);
              break; // 32MB
          case 8: control->dictSize = (1 << 26);
              break; // 64MB
          case 9: control->dictSize = (1 << 27);
              break; // 128MB -- this is maximum for 32 bits
          default: control->dictSize = (1 << 24);
              break; // 16MB -- should never reach here
          }
      control->overhead = (control->dictSize * 23 / 2) + (6 * 1024 * 1024) + 16384;

[pete4abw]

So many changes in testing. Withdrawing. See my fork for SDK 19 for current work.