Sanitize OTEL datastream

[rubvs]

Fixes elastic/apm-server#14043

@carsonip I am not exactly sure I should remove disallowed runes, or replace them with _.

[carsonip]

@carsonip I am not exactly sure I should remove disallowed runes, or replace them with _.

Replacing is fine, to make it consistent with ES reroute processor implementation: https://github.com/elastic/elasticsearch/blob/33d63fa9c2fc3afb9184f31e2681942cfd8dc7fd/modules/ingest-common/src/main/java/org/elasticsearch/ingest/common/RerouteProcessor.java#L203-L206

[carsonip]

After discussing with @felixbarny , we agreed that it will be easier to push for a change to loosen up the ecs spec such that namespace can contain -, which is how es reroute processor is implemented, rather than to push a breaking change in es. Therefore, can you change the implementation in this PR to allow - in namespace?

[rubvs]

After discussing with @felixbarny , we agreed that it will be easier to push for a change to loosen up the ecs spec such that namespace can contain -, which is how es reroute processor is implemented, rather than to push a breaking change in es. Therefore, can you change the implementation in this PR to allow - in namespace?

Just to make sure, the - is still disallowed in dataset?

[carsonip]

Just to make sure, the - is still disallowed in dataset?

Correct. Otherwise parsing dataset and namespace would be ambiguous

[carsonip]

lgtm, thanks for bearing with me!