It took a ten-year project to get a stabilized version that could be injected,

The technical knowledge of which I would be in possession of. And therefore know how to shortcut to the end product. Three or four rounds of experimental trials, each taking a couple of weeks, followed by intentional use of radical selection pressure to generate a strain that has the desired properties. Given the generational periodicity of 1 week per round, six months at most would be sufficient to produce a working product. (Also; deep tank fermentation was a large part of why pennicillin would be available in bulk to me. A larger problem would be the probenecid production, but that's a different story.)

Also -- the fact that pennicillin wasn't available in our history until 1952 in pill form wouldn't be much of an impediment nor even really an interesting question.

Also, windmills and waterwheels already existed.

Windmills, yes. In the hands of Persians, not in Romans. And they weren't moving quantities of water -- classical mechanics as we know it today was not 'invented' until over a thousand years later.

And where are you going to get an insecticide from, anyway?

There are many, many plants which produce insecticidal/nematocidal/fungicidal chemicals as a part of their normal lifecycles. Harvesting and processing them for these purposes is not exactly a technically complicated process. It just requires knowledge of them.