Clarify pattern matching is case-sensitive

[trask]

Related to open-telemetry/opentelemetry-java#8137

cc @SylvainJuge

[SylvainJuge]

To me, what I described in open-telemetry/opentelemetry-java#8137 is an inconsistency, which may come from the lack of a clear specification in this repository.

What should be done is either:

• "fix" the specification to make the case-sensitivity explicit, then modify the implementations and fix issues like IncludeExcludePredicate case sensitivity inconsistency opentelemetry-java#8137 in Java (1).
• consider this as part of the spec and keep things as they are (2).

This PR documents the current state and thus addresses (2), which would make the Java implementation aligned with the spec short term.

However, I think we should try to fix the underlying issue with (1) by removing ambiguity in the spec and settle on a consistent way to deal with case-sensitivity.

[trask]

What should be done is either:

• "fix" the specification to make the case-sensitivity explicit, then modify the implementations and fix issues like IncludeExcludePredicate case sensitivity inconsistency opentelemetry-java#8137 in Java (1).
• consider this as part of the spec and keep things as they are (2).

This PR documents the current state and thus addresses (2), which would make the Java implementation aligned with the spec short term.

However, I think we should try to fix the underlying issue with (1) by removing ambiguity in the spec and settle on a consistent way to deal with case-sensitivity.

@SylvainJuge I'm probably missing something, I thought this PR is fixing the specification to make the case-sensitivity explicit?

[SylvainJuge]

My bad, I completely mis-read the diff here, sorry for the noise.

I thought it only applied to "wildcard matching" whereas it also applies to the "exact match", so this aligns on having everything case-sensitive.

[zeitlinger]

Looks good — making case-sensitivity explicit is the right call. Consistent with how wildcards already behave, and SylvainJuge's concern is resolved in the comments.

[jack-berg]

Any comments on this @open-telemetry/configuration-approvers? Given other conversations about case sensitivity in JSON schema (like this conversation where we made a decision to make enum values lower_snake_case because JSON schema doesn't support case insensitive enums), I think we should adopt a stance of "case sensitivity is assumed unless explicitly stated otherwise".