Is there any architectural limitation in temporal with the continue as new functionality?
For the sake of the example, can i “continue as new” all my currently running workflows on demand maybe twice three times a day? (Workflow # is in the order of thousands)
Yes, absolutely, you can. The Temporal cluster scalability is limited by the overall number of operations per second it performs. Continue-as-new consumes some resources, so your cluster should be able to handle those. So if you call continue-as-new every second, then you can end up creating too much load in some cases.
A few thousand workflows calling continue-as-new a few times a day would not be a problem.