Google does not let you scrape.

That doesn't mean that you cannot still write a bot to scrape the content. If it's displayable in a web browser, then it's also parsable from Perl (or whatever language you choose).

Adding delays between page requests and/or distributing the requests would work around a lot of bot detection systems but in worst case scenario, there are projects out there that can solve Captcha's.

Essentially this can be viewed like music of movie piracy; if it can be played then it can be copied.

Sure, but throttling is going to make it really slow to gather all the data you want.

If you're stalking someone then you already have enough data to perform a more targeted scrape (ie what town to look up businesses from)

Laws do not allow employers to use information about protected characteristics to discriminate against employees (or potential employees) yet people are rightly worried about revealing that information.

People scrape Google all of the time, every day. All day long.