Incorrect header detection in CSV parsing 5.4.1

[lucas-inc-query]

I have encountered an issue with version 5.4.1 when parsing CSV headers. It seems that headers are being incorrectly parsed and appended with "_x".

To reproduce the issue, go to https://www.papaparse.com/demo, check the "Header row" box and parse the following string:

"foo,dog","foo,cat"

Inspecting the results > meta > fields property shows us the following array:

fields: [ "foo,dog", 'foo_1,cat"' ]

The 2 error messages are:

message: "Trailing quote on quoted field is malformed"
message: "Quoted field unterminated"

It seems that commas within the headers leads to the parser believing there duplicate header fields? This problem does not occur if the header: true option is omitted from the config.

[faulpeltz]

Unfortunately this seems to be the case with all quoted duplicate headers:
"bla","duplicate","duplicate","foo"

Results in:
fields: [ "bla", "duplicate", "duplicate\"_1,\"foo"]

• It works correctly when the quotes are removed
• Reverting to 5.3.2 also works (but has has no duplicate header renaming)

Edit:
As a follow up, looking at the code, the header duplicate detection code completely ignores quoting (just doing a separator split)

[basememara]

Related to #998