A timer is a tool for a specific kind of practice. Used by default, it quietly changes what the game is.
Timers are useful and overused. A timer turns a puzzle into a performance, and performance is a different activity from practice, with different mistakes. The question is not whether to time your rounds but which rounds deserve a clock.
A timer is right for one thing: measuring execution. If the route is findable and the question is how quickly you can take it, a clock makes the answer visible. That is a real skill, and it is worth training occasionally, especially if you find yourself losing rounds to hesitation rather than to search.
A timer is wrong for search and recovery. The moments where the game actually gets hard are the moments where you need permission to be slow: the middle of a thin start, the junction where two plans are equally plausible, the sudden wall. A clock on those rounds encourages a quick wrong move over a slow right one, which trains exactly the wrong reflex.
The practical arrangement is to keep most rounds untimed and use a timer for one session a week. That gives the execution practice without making speed the default. It also makes the timed session more useful, because the comparison is against ordinary rounds rather than against a permanently rushed baseline.
There is a second, softer use for a timer: a session limit. Set ten minutes for the whole session rather than for a round, and let the timer decide when you stop. That is a different tool with a different purpose, and it protects the habit on a busy day.
Try one week of untimed rounds and notice how the mistakes change. If the rounds get slower and the routes get shorter, the timer was making you pay for speed. If you want an unlimited stream of boards to use either way, the unlimited mode is the place; the daily round stays a single attempt.