cli: fix table in Tor anomalous measurements

[agrabeli]

I noticed that in some anomalous Tor measurements, the table (which displays an overview of results based on the testing of Tor directory authorities and Tor bridges) is inaccurate -- in comparison to the raw data.

For example, look at this measurement: https://explorer.ooni.org/measurement/20210416T000625Z_tor_AZ_39232_n1_cLbrywxhrppEuk7l

It says that connections to only 3/10 Tor directory authorites were successful (and this is corroborated in the raw data), but the UI of the table includes a tick next to each tested directory authority.

See:

This seems like a bug. We therefore need to fix the table so that it accurately displays which connections to Tor directory authorities (and bridges) failed.

[sarathms]

@bassosimone I am using test_keys.targets[${address}].summary.connect.failure (for the 'Connect' column). And then if (that.failure === null) I stamp that target with ✅ or if it has a string then ❌. If this is a wrong interpretation, what is a right one?

[bassosimone]

We looked into this with @sarathms and this is a bug in probe-cli, so we moved the issue into the probe repository.

[sarathms]

The particular target that has a clearer issue in the sample measurement is 45.66.33.45:443

[bassosimone]

So, on re-reading this issue, I don't think this is a probe bug, rather it's an UI bug that we should figure out how to address. The column is named "connect" and the tick means that we could connect. The error presented above says that the TLS handshake failed with a "generic timeout error".

What I find wrong about the table generated by https://explorer.ooni.org/measurement/20210416T000625Z_tor_AZ_39232_n1_cLbrywxhrppEuk7l is that there is no information about the handshake in all cases, which does not strike me as correct.