Added omit_requirement_level option for markdown table rendering

[AlexanderWert]

When rendering attributes tables outside specific semantic conventions (e.g. in an attributes dictionary), we need a way to omit rendering the requirement_level column in the table.

See this proposal:

• Proposal: Decoupling Attribute definition from Attribute usage in models semantic-conventions#197

This PR introduces an optional omit_requirement_level parameter in the markdown semconv_selector syntax that indicates the table rendering logic to omit rendering the requirement_level column in the corresponding attributes table.

[Oberon00]

I dont't think this should be a markdown-only option. Code generators would still see the requirement levels. IMHO, this needs to be somehow modelled in the YAML.

[AlexanderWert]

IMHO, this needs to be somehow modelled in the YAML.

I agree with that! But it sounds like a huge refactoring, so I was wondering if we could do this pragmatic step as a intermediate step to get open-telemetry/semantic-conventions#197 started before changing the yaml syntax significantly (so, as part of step 1 in open-telemetry/semantic-conventions#197 (comment)).

I dont't think this should be a markdown-only option. Code generators would still see the requirement levels.

Just curious: are code generators using the requirement level at all? IIUC, they are generating the language-specific constants for the attributes (which do not contain the requirement level in any form). Am I missing something?

[AlexanderWert]

IMHO, this needs to be somehow modelled in the YAML.

@Oberon00 See this initial proposal: #191

[joaopgrassi]

Pragmatically, I'd vote to get this intermediate step in so we can "unblock" open-telemetry/semantic-conventions#208. Later once we have defined the proper yaml in #208, we can simply revert the changes from here. I think this "intermediate" step makes sense.

[Oberon00]

Approved because the code looks fine and I don't want to block this. I have no opinion on whether we want to take this intermediate step or should try for a proper solution from the start.

[joaopgrassi]

I created #202 so we don't lose track to properly implement the final solution in YAML.

[joaopgrassi]

Can we have some tests just to make sure things are working as expected? We should have examples there that can make it easier (I think).

[AlexanderWert]

Can we have some tests just to make sure things are working as expected? We should have examples there that can make it easier (I think).

Will add some

[AlexanderWert]

@joaopgrassi I added a test for the table rendering while omitting the requirement-level column.

I think we are good to merge this and then do the release of the build-tools?