In short:
1. Underperforming workers can still be fired. Their ownership stake is bought out by the firm. Because firings are costly, hiring is a more intentional process, typically involving a "trial period" at the beginning of employment that more closely resembles traditional employment (no ownership stake, no voting rights, easy severance).
2. Successful cooperatives exist in all sectors, but there are many variables at play. A tech worker co-op today is unlikely to be able to pay its worker-owners salaries comparable to what VC can provide, due to factors ranging from tax rates (capital is taxed at negligible rates compared to labor) to investment bubbles and concentration of wealth. Long-lasting cooperatives have lasted longest in agriculture and finance because they can operate more efficiently than firms which need to retain a chunk of their margin to pay investors.
3. There is no real role for a union in a worker co-op. Surprisingly, this doesn't mean they don't exist. Mondragón actually went to war against unionizing efforts at their firm in the 80s, which led to them increasing education on what rights and privileges worker-owners already have. Generally, though, worker co-ops align the incentives of workers with the firm such that unions are unnecessary.
4. The management structure of each cooperative is up to them. Most successful ones (including Mondragón) delegate decision-making to a core group of elected executives, just like in successful modern democracies. Major structural decisions are made by general assembly -- analogous to a stockholder meeting at joint stock firms.
5. Worker co-ops differ from ESOPs (firms that grant lots of equity to workers) in that they confer decision-making, as well as equity, to workers. ESOPs usually grant workers non-voting shares. I've actually seen the ugly side of ESOPs when my startup employer diluted my stake to nearly nothing just prior to a buyout. That would be a much more difficult prospect in a worker co-op.
6. I'm unaware of any systematic studies done on the relative propensity for different types of firms to externalize costs. I'd only say that there's certainly less incentive for worker co-ops to act irresponsibly, as they are directly answerable to their community, who both work at and own the firm, and usually live nearby.
And there's no obvious way in which they seem more incentivised to bad behavior than other types of firms.
For example, Mondragón were "caught" underpaying some workers in their overseas factories, in line with how other firms abuse their outsourced workforce. However, they were caught by their own worker-owners, and unlike other companies, Mondragón's worker-owners objected to this treatment and voted to provide their outsourced employees with a path to full ownership.
7. I've responded elsewhere with what's holding worker co-ops back and won't elaborate too much here. Suffice it to say that they would be much more common if there was a standard corporate form, consistent and non-discriminatory tax treatment, and and easier access to alternative means of raising capital.
Where some or all of these structures are in place, cooperatives are surprisingly common. I'd advise looking into the Emilia-Romagna region of Italy, in addition to Basque Spain for more examples.