tools: perf: store metadata in the benchmarking results file

[janekmi]

Currently, when benchmarking is done a few JSON files come to existence e.g.:

.
├── apm_dram_vs_pmem.json
├── apm_pmem_vs_gpspm_pmem.json
├── ib_read_vs_rpma_read_dram.json
├── mix_rand.json
├── mix_seq.json
├── report.json
└── rpma_read_dram_vs_pmem.json

They are tailored to be input data for generating figures and the report but they also may be consumed using different measures. In doing so, the end-user may lose reference where given files come from and the result of what benchmarking they are, which is stored in other JSON files.

The idea is to store some metadata in those files just in case they would be decoupled from the rest of the resources.