Define p value type in the field label of the data file

[jdhayhurst]

As with effect statistic (beta, odds_ratio, hazard_ratio), could define the p value type (standard or -log10) in the field label itself instead of, or as well as, in the metadata. This would be consistent with the effect statistic handling and also allows users to interpret the data without reading the metadata file.

[marcora]

Another reason why separating sumstats from metadata in two different files is a weakness of the GWAS-SSF format in my opinion (in contrast to GWAS-VCF for example), particularly as it pertains to data integrity and self-contained documentation.