Gerard is far more opposed to macrolanguage projects than most of the other members here. Consistent with the way Amir put it, we need to be careful just how much we generalize on this topic: We "don't" want macrolanguages, but we also "don't" want projects in languages that are extremely close to each other, such that they're really mutually intelligible.  In some such cases, using the macrolanguage is going to be the most expedient approach, both linguistically and politically.

In this particular case, the test project for Marwari is coded with the macrolanguage code (mwr). But as it turns out, the principle constituent language of the macrolanguage is also called "Marwari", albeit with codes rwr (in India, in Devanagari) or mve (in Pakistan, in Perso-Arabic—which Ethnologue says "may or may not be the same as [rwr]"). There are also some related languages within the macrolanguage, some of which have very similar names (e.g., "Merwari", "Mewari").  

I'll try to confirm with the one current contributor, but it's entirely possible that this test is entirely in the constituent language Marwari–rwr; in that case, I can change the langcode in the request and mark it eligible. But that said, I wonder if it's really better to do that, or better to let the test continue using the macrolanguage code.

Steven