Bug 32335770 affects all your solutions that reference BINARY_AI. I suspect Oracle Text may be affected as well, though one would need to check. I do not know/remember whether Oracle Text uses the same base letter mapping table as BINARY_AI.
With compatible>=12.2.0 and max_string_size=extended, you can define the column as BINARY_AI. However, this will use this collation for all SQL comparisons of the column values, unless overridden. Otherwise, since Oracle Database 12.2, without the parameters set, you can just say:
name COLLATE BINARY_AI = input_name
in those specific conditions in which you really want accent-insensitive comparison. For now, bug 32335770 will prevent some specific characters to compare properly, though most should work fine. We are working on finding a solution to this bug. To work around it, you would need to explicitly map problematic characters to their non-accented counterparts using the TRANSLATE function and then compare with COLLATE BINARY_AI. Note, the bug specifically talks about LATIN CAPITAL LETTER L WITH STROKE, which is used in Polish only, but we have not yet checked whether more characters are affected.
Note that by design, the BINARY_AI collation does not equivalence certain letters that you would normally want to be compared as equal. For example, it will not compare German eszet 'ß' as 'ss' and it will not compare the ligature 'œ' as 'oe'. For this, you may need to use one of the more expensive but more versatile collations such as GENERIC_M_AI or UCA0700_ORAROOT_AI.