Test metadata makes results unreadable

[grdumas]

When reviewing a result CSV (intrate or fprate), each component result of the benchmark is preceded by 12 lines of metadata. For example the intrate results file includes 10 component results (500.perlbench_r, 502.gcc_r, 505.mcf_r, 520.omnetpp_r, 523.xalancbmk_r, 525.x264_r, 531.deepsjeng_r, 541.leela_r, 548.exchange2_r, 557.xz_r), which are buried within 120 lines of duplicate metadata.

Having the metadata once at the top of the CSV file is fine, but as currently stands, the repetition of metadata makes results files much harder to read.