D002_02_01_inc: 204 kept (a source's 404 is no error - a wide registration's...

*** Test Cases ***

D002_02_01_inc Delete Entity On The Context Broker

    [Documentation]    Verify that, when one has an inclusive registration on a Context Broker, one is able to delete entities locally and should get a BatchOperationResult structure

    [Documentation]    Verify that, when one has an inclusive registration on a Context Broker, one is able to delete an entity that exists locally. A 404 from an inclusive Context Source that simply does not hold the entity is not an error (TS 104-175 § 10.2.8.4 reserves ResourceNotFound for the case where neither the broker nor any registration knows the entity), so the operation returns 204 No Content.

    [Tags]    since_v1.6.1    dist-ops    4_3_3    cf_04    additive-inclusive    4_3_6_2    5_6_6    6_3_3

    Set Stub Reply    DELETE    /ngsi-ld/v1/entities/${entity_id}    404

    ${response}=    Delete Entity    ${entity_id}

    Check Response Status Code    207    ${response.status_code}

    Check JSON Value In Response Body    ['status']    404    ${response.json()['errors'][0]['error']}

    Check Response Status Code    204    ${response.status_code}

    ${response_retrieve}=    Retrieve Entity    ${entity_id}    local=true

    Check Response Status Code    404    ${response_retrieve.status_code}

**Upstream (2026-10-03):** spec question - forge #105; provisional d5ed03c3.

**Update (2026-10-08) - REVERSED for D002_02_01_inc:** "the spec is silent" below is wrong. TS 104-176

§ 6.3.5 exempts a source's 404 as "not abnormal" for distributed **GET** only; "In the case of an

inclusive or redirect registration, … when updating the state of the distributed entity, an error

response is returned from one or more registered sources: 207 Multi Status" (also § 7.3.3.4). Develop's

207 is right; this branch takes it back, and coraine reports the source's 404 (207). D002_02_02_inc is

unaffected (the source deletes successfully). The idempotency argument goes to the spec instead

(spec-doubts #91). Raised by geolonia#32.

**Update (2026-10-08) - our position, raised with ETSI (spec-doubts #91):** "the spec is silent" below

is wrong - TS 104-176 § 6.3.5 exempts a source's 404 as "not abnormal" for distributed **GET** only, and

for an inclusive or redirect registration "when updating the state of the distributed entity, an error

response is returned from one or more registered sources: 207 Multi Status" (§ 7.3.3.4 the same), which

develop's test follows. We hold that the exemption was simply never extended to the other verbs: with a

registration on a **type** (no id), every source that does not hold entity X answers 404 to a forwarded

PATCH / DELETE / … - the normal case of a wide registration, not an error - and the operation, done

where X is, succeeded: 204. This branch keeps 204 (coraine answers 204) until ETSI decides.

**Hit:** Both tests do `Delete Entity` against an inclusive CSR

where one of the two legs (local or remote) is supposed to