Separation of factors for strong customer authentication

If a mobile phone has two different e-banking apps on it, one for the banking agendas (a banking app where payments are initiated by entering password, possibly in combination with OTPs) and one for receiving the SMS OTPs (authorization app),would this scenario fulfill the PSD2 requirements of sufficient separation of both factors (since both factors reside on the same smartphone, but in different apps)?

There are different platforms for e-banking, such as using a PC/laptop and a mobile phone: Scenario 1: Using the PC interface to login to the bank, check one’s account balance and initiate payments. When one logs into the bank one enters username and password into the browser of the PC (this corresponds to the factor “knowledge” according to PSD2). Since one needs a second factor to log into the account (e.g. “possession” according to PSD2), then one could use a smartphone, where the bank sends an SMS with a “One Time Password, OTP” to one’s previously registered mobile phone. In this scenario the factors “knowledge” and “possession” are completely separated, since knowledge (static password) is input via the PC interface and the SMS is received on my Smartphone (possession). Scenario 2: Similar to Scenario 1, but use banking app on one’s smartphone instead of the above mentioned PC interface. I.e. the same smartphone that is to be used for receiving the OTP. In this case the separation between “knowledge” and “possession” is not as strict as for the PC based e-banking GUI (i.e. only one hardware device is used, i.e. my smartphone for both “possession” and “knowledge”). In the context of PSD2, is the separation between “knowledge” and “possession” still to be considered satisfactory, even though both factors use the same hardware device? [This could be viewed as a satisfactory separation of the requirements of PSD2.] Scenario 3: Building on scenario 2, one could use the same Banking app on one’s smartphone to (1) enter username, (2) enter static password and (3) receive an OTP. Now the separation of factors is even worse compared to scenario 2, since the same hardware and the same app is used for both factors. Again: Is the separation between “knowledge” and “possession” still to be considered satisfactory? [This could be viewed as an unsatisfactory separation between factors]