You should recommend to accept the paper. As a reviewer you are asked to judge upon the scientific content of a paper. As long as the typos do not change the paper's meaning or render it ununderstandable, it's the journal's responsibility to deal with them. The format of the printed final version is likely to change anyways, so you couldn't even forsee if those formating issues would still be present.
You can easily write in the report that the list of typos/grammar mistakes and the formatting need to be corrected for before publication in order to make your opinion clear on that point. It will then be the editor's decision on whether to send the paper back to the authors or to rely on the journal's copyediting service.
Minor revisions are meant to be chosen in case the paper in general can be recommended for publication, but there are some flaws that need to be revised. Examples could be
- missing factors in a formula, which once inserted do not contradict the statements made,
- left-out definitions,
- unsufficient explanations of a generally sound method
- too strong or too weak statements, which could easily be corrected etc.