Is it ethical for a journal to cancel an accepted review request when they have obtained sufficient number of reviews to make a decision?

I think that in this particular case, the journal behaved ethically. A deadline is a contract, and if one of the involved parties breaks that contract (notably, without seeking communication), it's up to the other party to decide how to deal with that.

Cancelling a solicited review is as much "unethical" as not keeping the deadline without noticing the editor.

However, I would say this more as a question of

good manners instead of ethics.

Keep in mind that in most cases the solicitation and handling of reviews is in fact communication between scientists. You do not really communicate "with a journal" or "with a company", but more with a colleague. While this is often obfuscated by software and automatically generated emails, there is usually a colleague sitting on the other side, asking you for a review within some time frame. Editors often strive (and sometimes get pressure) to drive the review time down and hence, try to insist on their time frame. Somehow loosely speaking: "If you don't bother to reply if you can't make the deadline, why should they bother to wait for your review?" In case you can't make the deadline, inform the editor and ask for extension. That is the way to go - it's polite and everybody knows what is going on.

For example, if you write "I can't meet the deadline, but I am halfway through to the paper and will have my report on…", the editor will most like extend the deadline. If you, however, write "I can't meet the deadline, and I haven't started to review yet." it may well be that the editor answers, that your report is not longer necessary (although, I do not expect this to happen too often).

To the edit: I think you have done the exact right thing. Also this shows that politeness and respect are central to the conversation between editor and reviewer. Both do an extremely valuable, important and time consuming job for the scientific community and know that of each other. A reasonable question asked a the right time usually gets a reasonable answer and one email spend to clear up an issue is usually worth it.

Actually, for this result to happen, it is not necessarily the case that the journal "solicited more reviewers than what is necessary".

I know that there are journals that request two referee reports, but as soon as one negative report arrives, the paper is rejected and the other referee report is cancelled.