Can I have two named connections of the same integration at once? Yes — each with its own name, its own credentials and its own allowed actions. A flow piece picks its own by name in the “Connection” field (Manual 55, section 3.5).
What happens if I rename or revoke a connection a flow is using? Revoking it disables the flows that were using it (the notice tells you which ones); renaming it doesn’t break anything, the flow still identifies it by its internal id, not by the name you see on screen.
Why doesn’t “Email (read)” have a “main” connection? Because a single mailbox shared by default for the whole tenant doesn’t make sense — every mailbox you connect is always an explicit choice, with its own scope.
Can I write to any sheet or folder the connected account can see? No — only to the ones you’ve explicitly registered as a resource with write permission (section 3). It’s a separate lock from the connection’s “allowed actions”, meant so that enabling “Add row” never opens the door to writing everywhere else.
Is the “Invoices by email” template an end-to-end automation? Yes — it chains reading the email, extracting the data, validating it and saving it without anyone having to copy anything by hand, and it only asks you to approve when confidence is low or something doesn’t check out. It’s tested end to end against a real mailbox (section 5); the recommendation is to activate it first against a test mailbox or a scoped folder and review the first jobs before leaving it unsupervised — the first real go-live against your production mailbox is done with support, not because anything is left unfinished.
Where are a flow’s pieces and generated inspector configured? That’s the visual editor (builder) — Manual 55, section 3.