Fix older cEOS version empty ping failure validation

[jbemmel]

Fixes #3496

[ipspace]

In principle, I hate slapping random fixes into the code base to address niche issues, and this seems like one of those.

First, what "older version" are we talking about? I had no problems for ages, and I'm not particularly fast at upgrading cEOS. Also, what's wrong with downloading not-so-ancient version from Arista's website?

Also, the netlab validate tests execute Docker or SSH commands, and examine the outputs only when those commands are completed. So I have to wonder under what scenario ping executed as an SSH command (that's what we're using on cEOS) returns no output.

[jbemmel]

First, what "older version" are we talking about? I had no problems for ages, and I'm not particularly fast at upgrading cEOS. Also, what's wrong with downloading not-so-ancient version from Arista's website?

I used harsha258/ceos:4.32.0F

I cannot download from the Arista website because I do not have a "corporate address" and @gmail.com isn't accepted

As documented in #3496: In vrf/12-multi-vrf-bgp.yml, the inter-VRF isolation validation using clab.image: harsha258/ceos:4.32.0F reported:

[inter_vrf] Ping-based isolation test between blue and red VRF [ node(s): r1 ]
[FAIL] Node r1: Ping to r4 did not fail
Input data: {"stdout": ""}
Plugin expression: ping('r4',expect='fail')

[ipspace]

OK, didn't take me long to figure out what the problem was: that particular version of cEOS prints the "ping: connect: Network is unreachable" to stderr, not stdout. Maybe we should be fixing the root cause instead?

[ipspace]

Replaced by #3502 (this one is dancing around the symptom, not fixing the root cause)