I think the use-cases and requirements differ quite a bit between radio stations and individual musicians/bands. While there'd no doubt be plenty of overlap (audio player, social media integration, etc.) radio stations usually have full-time staff, advertisers, and so forth. Hosting a front-end for stations is thus likely to require much more robust SLAs, which can thus entail higher prices. Of course, there are additional costs involved as well.
The "physical" part of hosting should never amount to a significant portion of the costs, as it's pretty much a commodity, and competition is driving prices down. The HR costs for support, maintenance and development are likely to be much higher.
In my experience, clients are much more likely to finance the development of new features, as opposed to general maintenance, such as adding/improving documentation, tests and performance, bug fixing, and refactoring (i.e., reducing technical debt). This is why I've insisted on rigorous documentation and testing be built into the development process from the very beginning. That up-front investment makes long-term support much more efficient.
AWS' free usage tier covers only a the first year of a micro instance, as far as VPSs go. Otherwise, their pricing tends to be on the high side, as they offer a premium service. I think Linode would be a good, lower-cost alternative
1) how much time per wk/month could you foresee putting into this effort;
On an ongoing basis, I don't see being able to work more than ~10 hrs/wk, considering my existing commitments. For short-term bursts, say to launch a client project, or the like, I could put in 20-30 for a week or two at a time. That said, I'm willing to work at $100/hr, instead of my usual $150/hr rate.
2) with regards to the roles outlined, would it be ideal to have a few goto people who could commit a regular amount of time to act as leads?
I'd re-organized that page to be more task-oriented. But it's hard to say more at this point. We have a number of goals outlined, culminating in a reasonably feature-complete, polished SaaS product. Working backwards from that, we'll need an MVP of the SaaS product that provided some tangible value. Before that, we'll need a framework, to build out some bespoke solutions that can form the basis of the MVP.
If we have clients with budgets for bespoke solutions (understanding that they're seeding OpenProducer) then I work with subcontrctors who could fulfill PM/analysis/QA, UX/theming, and developer roles respectively.
As for the OpenProducer Project, in the FLOSS sense, I think leadership in various roles is something that emerges from community dynamics. It needs cultivation, rather than hiring.
3) knowing my desired role/time commitment (and yours), what roles do we need to fill if we're going to first focus on building sites for stations using the D7 platform we already have.
I'd try to group the tasks as above: PM/analysis/QA, and UX/theming. Depending on the scope of the projects, the complexity of customizations, upstream bug-fixes and features, etc. we will likely need additional developers/programmers as well.