implement max_mem_mb option in the config file to resolve memory issues withsample_level_ibd rule

[jaamarks]

With eb116a5 we added the following code to the sample_level_ibd rule in the sample_qc.smk file so that we could apply the max_time_hr workflow parameter to this rule:

        mem_mb=lambda wildcards, attempt, input: max((attempt + 1) * input.size_mb, 1024),
        time_hr=lambda wildcards, attempt: BIG_TIME[attempt],

But now we are experiencing issues with the mem_mb not being aggressive enough. A sample size of 110K quickly failed on multiple attempts because of memory issues. So we need to make this more aggressive.

[jaamarks]

The previous allocation was set in the genome rule in the plink.smk file at:
PLINK_BIG_MEM = {1: 1024 * 4, 2: 1024 * 64, 3: 1024 * 250}

From looking at log files of previous successful runs of a 110K sample, the pipeline failed up until it allocated 256GB at which point it succeeded. The pipeline would fail very quickly (under a minute) if it was a memory issue.

[jaamarks]

We could implemented a max_mem_mb parameter in the config, and then apply the following approach for any rule that requires a significant amount of memory (like the sample_level_ibdrule):

PLINK_BIG_MEM = {1: max_mem_mb / 3, 2: max_mem_mb / 2, 3: max_mem_mb}