docs404 Test Fails when Hugo Not Running

Nobody has claimed this yet.

The script `ci/scripts/docs404.sh` doesn't check whether Hugo had been running prior to actual test. This cause Travis CI test fails on random PRs, with symptoms logged like (taken from [Travis log](https://travis-ci.org/linode/docs/jobs/612238062?utm_medium=notification&utm_source=github_status) for #2891):
```
2019-11-15 06:28:11 [scrapy.extensions.logstats] INFO: Crawled 0 pages (at 0 pages/min), scraped 0 items (at 0 items/min)

2019-11-15 06:28:11 [scrapy.extensions.telnet] DEBUG: Telnet console listening on 127.0.0.1:6023

2019-11-15 06:28:11 [scrapy.downloadermiddlewares.retry] DEBUG: Retrying (failed 1 times): Connection was refused by other side: 111: Connection refused.

2019-11-15 06:28:11 [scrapy.downloadermiddlewares.retry] DEBUG: Retrying (failed 1 times): Connection was refused by other side: 111: Connection refused.

2019-11-15 06:28:11 [scrapy.downloadermiddlewares.retry] DEBUG: Retrying (failed 2 times): Connection was refused by other side: 111: Connection refused.

2019-11-15 06:28:11 [scrapy.downloadermiddlewares.retry] DEBUG: Retrying (failed 2 times): Connection was refused by other side: 111: Connection refused.

2019-11-15 06:28:12 [scrapy.downloadermiddlewares.retry] DEBUG: Gave up retrying (failed 3 times): Connection was refused by other side: 111: Connection refused.

2019-11-15 06:28:12 [scrapy.downloadermiddlewares.retry] DEBUG: Gave up retrying (failed 3 times): Connection was refused by other side: 111: Connection refused.

2019-11-15 06:28:12 [scrapy.core.scraper] ERROR: Error downloading

Traceback (most recent call last):

File "/home/travis/virtualenv/python3.6.7/lib/python3.6/site-packages/scrapy/core/downloader/middleware.py", line 43, in process_request

defer.returnValue((yield download_func(request=request,spider=spider)))

twisted.internet.error.ConnectionRefusedError: Connection was refused by other side: 111: Connection refused.

2019-11-15 06:28:12 [scrapy.core.scraper] ERROR: Error downloading

Traceback (most recent call last):

File "/home/travis/virtualenv/python3.6.7/lib/python3.6/site-packages/scrapy/core/downloader/middleware.py", line 43, in process_request

defer.returnValue((yield download_func(request=request,spider=spider)))

twisted.internet.error.ConnectionRefusedError: Connection was refused by other side: 111: Connection refused.

2019-11-15 06:28:12 [scrapy.statscollectors] INFO: Dumping Scrapy stats:

{'downloader/exception_count': 6,

'downloader/exception_type_count/twisted.internet.error.ConnectionRefusedError': 6,

'downloader/request_bytes': 1170,

'downloader/request_count': 6,

'downloader/request_method_count/GET': 6,

'finish_reason': 'finished',

'finish_time': datetime.datetime(2019, 11, 15, 6, 28, 12, 266530),

'log_count/DEBUG': 7,

'log_count/ERROR': 2,

'log_count/INFO': 2,

'memusage/max': 53145600,

'memusage/startup': 53145600,

'retry/count': 4,

'retry/max_reached': 2,

'retry/reason_count/twisted.internet.error.ConnectionRefusedError': 4,

'scheduler/dequeued': 6,

'scheduler/dequeued/memory': 6,

'scheduler/enqueued': 6,

'scheduler/enqueued/memory': 6,

'start_time': datetime.datetime(2019, 11, 15, 6, 28, 11, 76552)}

2019-11-15 06:28:12 [urllib3.connectionpool] DEBUG: Starting new HTTP connection (1): localhost:1313

Hugo server not running on port 1313

The command "./ci/scripts/$TEST_SUITE.sh" exited with 1.
```

Currently the workaround is pushing additional commits to PR, (hoping that) Hugo will run in the CI.

## Suggested Fix

Add check if Hugo was running. This ideally should be placed before `scrapy` invocation. Perhaps the check can be done by using `curl http://localhost:1313`. If `curl` return error, attempt to `hugo server`.

PS: Other means for checking Hugo runs are welcome.

Start with ci/scripts/docs404.sh and inspect the Hugo and scrapy invocation order. Reproduce the localhost:1313 failure, then use the issue's curl-based check as a starting point for ensuring Hugo is available before scrapy runs. Done means docs404 no longer fails solely because Hugo was not running.

Written by the indexing model from the issue text.