I expect the messages will be sent correctly, the problem is that they may not be received at the mega correctly. The errors that were reported with the 6n138 are that the rising edge is too slow (because it is being pulled up rather than driven up). That means that while the midi device probably sent the message correctly, the mega doesn’t always receive it correctly. Since the source of the midi messages is a script it should be fairly easy to test this. Modify the source script to send the same midi message something like 1000 (or even 10,000) times. On the mega create a program that reads incoming midi messages, compares the message to what should be sent (and counts an error if the message received doesn’t match what was sent, easy to do when always sending the same message, if it doesn’t match at the mega, then an error occurred in transmission) and counts correct messages. When the source has sent its 1000 (or 10,000) messages check the count on the mega and see if the good message count plus the error count match the number the script sent. It they don’t match then some messages were lost completely.This may take a long time, but you can run it overnight (or for a few days if needed) to see what the error rate looks like. This would also be a good tool to keep around, in case you run in to a situation where messages look to be being lost. If you run this program on the link where messages are being lost it should tell you if the errors are occurring on the serial link (as opposed to somewhere else in the system). If the mega doesn’t receive all the messages then you have the system error rate and you can decide if it is acceptable (a couple of losses in 1000 is perhaps acceptable, 500 losses in 1000, i.e a 50% error rate, probably is not). Doing the same test with the 6n137 installed is also worthwhile to see if it has a lower error rate.
Note Old_Grey’s comment on making changes when more than one view is routed. At worst you have the current version to fall back to if a change in pcb breaks the other views. I’d advise keeping a copy of working sketches as you proceed so you don’t lose a large amount of work to an error that corrupts the database. Start by positioning the parts in pcb view in such a way as to make routing easy. Unfortunatly it is trial and error (and autorouting works so poorly that it isn’t worth even trying for a board like this!). If you find that you can’t route some traces, see if moving one or more components around will make routing easier. It is hard and experience is the only way it gets any easier. You can also always post your current sketch and one of us can tell you how we would do what you want to do (as a first board that’s what I would do, one of us can probably make a suggestion that will save you a lot of trial and error). Once its done, before ordering boards post the final result here for some of us to check for you as it is easy to make mistakes that produce non working boards. Good luck!
Peter