feature request: bbmap recognize interleaved bam input

[eboyden]

It's nice that bbmap accepts bam input given that fastqs often do not and cannot include metadata that clinical pipelines in particular require like @PG header lines, or other data stored in tags. However it only aligns reads as though they were unpaired.

Is there a reason why the logic used for input fastqs can't also be applied to input ubams? I.e. if interleaved=t is specified, require sequential pairs of identical read names (with or without /1 and /2), exit with an error code otherwise. If interleaved=auto is specified, map pairs of identical read names as paired, map as single-end otherwise. If input is streamed, require interleaved=t.

Thanks!