Fix CHANGELOG.md (server.name does not exist)

[Oberon00]

Changelog introduced in open-telemetry/opentelemetry-specification#3402 refers to a nonexistent attribute.

[Oberon00]

Oh, it seems that the GH online editor insists on adding not one but two final newlines, and even creates empty commits to enforce that (53acf50)

[Oberon00]

BTW, the staments of the form "net.* only applied to client/server" are not true. The mentioned "non-applicable" cases may have been less commonly used or exotic or even never used, but purely from the spec / conventions point of view for the general attributes (not talking about specific contexts like http), peer and host were defined completely independent of any higher-level protocol's logical role like "client" or "server"

[arminru]

Suggested change

	`net.host.name` and `net.host.port` to `server.address` and `server.port` (since `net.host.*` attributes only applied to server instrumentation),
	`net.host.name` and `net.host.port` to `server.address` and `server.port` (since `net.host.*` attributes primarily applied to server instrumentation),

How about this then?

[Oberon00]

This is IMHO not enough of a reason and should be changed to the more wordy:

Suggested change

	`net.host.name` and `net.host.port` to `server.address` and `server.port` (since `net.host.*` attributes only applied to server instrumentation),
	`net.host.name` and `net.host.port` to `server.address` and `server.port` on server side (typical case) and `client.address` and `client.port` on client side,

But that would need to be changed in all the other points below as well.

It seems odd that a CHANGELOG entry produces so much food for discussion, how about removing the mapping from here entirely for now and instead doing what #43 suggests with a proper document?

[reyang]

Resolving for now per #126 (comment).

[Oberon00]

(unresolved after merge for easier future reference)

[arminru]

Suggested change

	`net.host.name` and `net.host.port` to `server.address` and `server.port` (since `net.host.*` attributes only applied to server instrumentation),
	`net.host.name` and `net.host.port` to `server.address` and `server.port` (since `net.host.*` attributes primarily applied to server instrumentation),

How about this then?

[Oberon00]

I suggest to merge this as-is, since it fixes an obvious editorial mistake, and any remaining changes regarding client vs. server side can/should be discussed in #43 (and possibly #62).

[reyang]

I suggest to merge this as-is, since it fixes an obvious editorial mistake, and any remaining changes regarding client vs. server side can/should be discussed in #43 (and possibly #62).

+1