Treat an unreadable backend probe as the older server, not an error

Stalwart before 0.16 does not know urn:stalwart:jmap, and rejects the whole
request rather than the one call when `using` names a capability it cannot
parse. The probe is only sent when the session advertises that capability, so
this should not arise — but if it ever does, throwing turns a server we can
still manage credentials on into a Security page that only shows an error.
Fall through to the endpoint those servers do have.

server/src/account.ts

@@ -83,11 +83,17 @@ async function probeBackend(ctx: Ctx): Promise<Backend> {
   // A server with the registry answers x:AccountPassword/get; one without it
   // fails to parse the method name at all and returns unknownMethod.
   if (ctx.session.capabilities && STALWART_CAP in ctx.session.capabilities) {
+    try {
       const res = await jmap(ctx, [["x:AccountPassword/get", { accountId: accountId(ctx), ids: [SINGLETON] }, "p"]]);
       const [name, args] = res.methodResponses?.[0] ?? [];
       if (name && name !== "error") return "registry";
       const type = (args as { type?: string } | undefined)?.type;
       if (type && type !== "unknownMethod") return "registry"; // present, but refused us
+    } catch {
+      // Not an answer we can read - most likely a server too old to know the
+      // capability we named, which rejects the whole request rather than the
+      // one call. Fall through and try the endpoint such servers do have.
+    }
   }
   return "legacy";
 }