Adding apache and nginx access log tests

[eyalkoren]

Adding tests to try and reproduce failures in ingest pipeline following #121914.

A benchmark ingesting bulks of messages from different sample sources fail specifically on the ingestion of some (apparently random) nginx apache access logs.

@gbanasiak found the following example log event causing the failure:

{"@timestamp": "2020-09-02T08:21:51.000Z", "type": "beats", "input": {"type": "log"}, "agent": {"version": "7.3.2", "id": "2e5b12d4-8f16-4bfe-8d91-4a98dd9c7214", "type": "filebeat", "name": "infosec-ci-master-green.c.elastic-ci-prod.internal", "ephemeral_id": "3f2b9397-a5e0-4c51-a814-5cce75962b97", "hostname": "infosec-ci-master-green"}, "service": {"type": "nginx"}, "host": {"os": {"family": "debian", "platform": "ubuntu", "kernel": "5.3.0-1032-gcp", "version": "18.04.3 LTS (Bionic Beaver)", "codename": "bionic", "name": "Ubuntu"}, "architecture": "x86_64", "containerized": false, "id": "bfafdfcc69fc1f239af7a05fe266e68c", "mac": ["42:01:0a:e0:01:ec"], "ip": ["10.224.1.236", "fe80::4001:aff:fee0:1ec"], "hostname": "infosec-ci-master-green", "name": "infosec-ci-master-green.c.elastic-ci-prod.internal"}, "event": {"timezone": "+00:00", "module": "nginx", "dataset": "nginx.access"}, "log": {"file": {"path": "/var/log/nginx/infosec-ci.elastic.co.access.log"}, "offset": 1040193}, "ecs": {"version": "1.0.1"}, "cloud": {"provider": "gcp", "instance": {"id": "7729978443851144062", "name": "infosec-ci-master-green"}, "machine": {"type": "n1-standard-4"}, "project": {"id": "elastic-ci-prod"}, "availability_zone": "us-central1-a"}, "@version": "1", "tags": ["jenkins_master", "nginx", "infra-stats", "\ud83c\udd71\ufe0f", "llama", "\ud83e\udd99", "llama-prod"], "fileset": {"name": "access"}, "message": "28.27.251.216 - dustin03 [03/Jan/2020:21:05:52 +0000] \"GET /computer/api/json HTTP/1.1\" 200 602 \"-\" \"Go-http-client/1.1\"", "data_stream": {"type": "logs", "namespace": "default", "dataset": "nginx.access"}, "rally": {"message_size": 120, "doc_size": 1531}

The error is:

Bulk request failed: [HTTP status: 400, message: [1:2460] failed to parse: data stream timestamp field [@timestamp] is missing]

This PR contains a test that attempts to reproduce the issue with the same pipeline as used in the failing benchmark and the same message that is causing the failure.
Passing this test message in isolation through the same pipeline doesn't cause any issue, thus the root cause is likely such that require more real world scenarios, like big bulks and high concurrency.