Captured from a real `junit-platform-console-standalone --details=tree` run against
CsvSourceTest#testPublishingPermissions, compiled and executed for real.

╷
├─ JUnit Platform Suite ✔
├─ JUnit Jupiter ✔
│  └─ CsvSourceTest ✔
│     └─ testPublishingPermissions(int, String, boolean) ✔
│        ├─ user USER_ID = "1001" with role ROLE = "EDITOR" ✔
│        ├─ user USER_ID = "1002" with role ROLE = "ADMIN" ✔
│        └─ user USER_ID = "1003" with role ROLE = "USER" ✔
└─ JUnit Vintage ✔

Test run finished after 228 ms
[         3 tests successful      ]
[         0 tests failed          ]

This is the genuinely new, easy-to-miss finding in this rewrite. The custom template is
name = "user {0} with role {1}" -- plain positional placeholders, the same {0}/{1} syntax used
everywhere else in this module. But because @CsvSource(useHeadersInDisplayName = true) is also
set, each individual {0} and {1} does not resolve to the bare value ("1001", "EDITOR") the way
it would without that flag -- it resolves to "HEADER = value" ("USER_ID = \"1001\"",
"ROLE = \"EDITOR\""). useHeadersInDisplayName changes what every positional placeholder expands
to, not just the auto-generated default name used when no custom name is given at all. Compare
this against docs/output/05-csv-source-headers-default-display-name.txt (no custom name, same
data, same flag) and docs/output/06-header-name-placeholder-throws.txt (an attempt to put the
header name directly inside the template, which is a different thing entirely and throws).