Part of the problem with the process is that, for various well-intentioned reasons, there's a major focus placed on delivering objective evaluation on any kind of funded research or work. Bureaucrats want to demonstrate the impact their funding had, so that next time around (in say 5 or 6 years) they can make the case for the same amount of money (or indeed more). This leads to a system where "safe pairs of hands" are favoured, due to their ability to "play the game" and deliver the right kinds of metrics. A larger organisation can deliver "more jobs created" as they have the operating capital in place to run hiring using the research funding money, and commit up-front to "creating" those jobs (which they would have needed anyway, but can now fund, in part or in whole, through the funded project).
That's why, in my view, the bigger companies and universities do well - they have a scale that helps them to deliver these kinds of outputs. As a small company, the "spend-and-claim-back" approach to most research funding can be a real issue, especially if the claims are delayed due to bureaucratic "checks and balances". These are necessary to prevent blatant outright fraud (i.e. people not doing the work they say they did, and pocketing the cash), but they tend to be applied across the board, rather than in a targeted way proprortional to the level of risk, and size of the organisation. The end result is smaller players spend more time (proportionately) handling bureaucracy if they do win funding.
RE selecting meaningful projects, this is arguably because EU research funding looks further ahead, at lower technology readiness levels. An open source project used by everyone is "high TRL" and therefore hard to fund. If they have an entity (as you point out, many don't, which makes them harder to fund),
If you don't have a company to pay yourself through an official payroll system, I believe there are rules in place which effectively define that you work a maximum of 8 hours per day, 5 days per week, and that you can't possibly earn more than 25 EUR per hour, which is of course completely out of line with the sector, and makes it really hard to work without creating a company. At which point you then become subject to state aid rules, and need to keep a handle on that.
And beyond the research programmes, any kind of "innovation funding" then drags you into the state aid system of partial funding, which is very difficult for a small company - getting 60% of full payroll costs funded sounds nice, but it requires you have the cashflow/capital in place to run payments through payroll, in order to claim back a percentage. Not ideal for the kinds of non-commercial "critical internet infrastructure" that need this kind of funding the most.