When creating a new serial subscription, it used to be that we could enter either a subscription length (number of issues, months or weeks) OR a subscription end date. When we would test the prediction pattern, it would show the right number of issues based on one of these values. Now it seems that if we only enter a number of issues for the subscription length and test the prediction pattern, not only does it tell us we're not being consistent with our end date (I thought it was optional?) but it also doesn't show the correct number of issues.

To test:
1. For this, I created my own record, but you can skip this step and use any existing record
   1.1. Go to Cataloging
   1.2. Click on New record
   1.3. Fill out the manadatory fields (in MARC21, typically 000, 003, 005, 008, 040$c, 245$a and 942$c)
   1.4. Click Save
   1.5. Note the record number, do not create an item

2. Create a serial subscription
   2.1. Go to Serials
   2.2. Click on New subscription
   2.3. Enter the record number
   2.4. Click Next (and confirm you're not using a vendor)
   2.5. Fill out the second form as follows
        - First issue publication date: 2023-01-01
        - Frequency: 1/month
        - Subscription length: issues 24
        - Subscription start date: 2023-01-01
        - Subscription end date: leave empty
        - Numbering pattern: Number
        - Locale: leave empty
        - Begins with: 42
        - Inner counter: leave empty
   2.6. Click on Test prediction pattern

   --> Note there is a message saying "End date is not consistent with subscription length." even if, normally, only one is needed

   --> Only the issues until today are displayed (in my case, only the January 2023 issue), even if we said our subscription was for 24 issues

There is no problem if we enter a subscription end date. This means either the subscription length problem is a bug, or end date is now mandatory and should be displayed in red like other mandatory fields.