add blosc tranform layer

[psychocoderHPC]

• add CMake integration
• add autotools integration
• update documentation
• add blosc transform support

transform options

---

• shuffle: default = no
  • no disable pre-conditioner
  • bit shuffle bits and bytes
  • byte shuffle bytes
• compressor: if blosc is compiled with all compression libraries, default = memcpy available compressors: zlib, lz4, lz4hc, snappy, zstd, blosclz, memcpy (plain data copy)
• lvl: valid range [1;9], lower number means faster compression but lower compression ratio, default = 1 (fast)
• threads: valid range [1;number of native threads], default = 1
• threshold: in byte, valid range [1;2^31], default = 128

scaling within a real word example PIConGPU

You can find more information about the importance of this transformation layer in the pre-print Axel Huebl, et al., On the Scalability of Data Reduction Techniques in Current and Upcoming HPC Systems from an Application Perspective, accepted for DRBSD-1 in conjunction with ISC'17, arXiv:1706.00522

CC-ing: @ax3l