A lot of road salt is brought into the parkade on vehicles during the winter months and obviously this takes a toll on the receptacles.
This Corrosive Environment might warrant the use of Enclosures and / or Device Covers with NEMA 4X ratings.
Besides the Corrosion (Redox Reaction) Problems associated with the road Salts, the Material makes an efficient Conductive "Bridge", allowing "Semi-Impeded" flow of Charges (Electrons and Holes) via Na and Cl Elements.
Basically, the Arc Fault You + Your DVM's Leads experienced during the Voltage Test, may have been exacerbated (or fully due to) the Road Salts concentrations within &/or surrounding the Device (Receptacle).
Nevertheless, the Receptacle in the posted Image is beyond trashed; the Box is filling with Washed-Down Road Salt, and it looks like any attempt to backout the 6-32 Mounting Screws is hopeless at best!
(exception: large chisel to remove the screw heads...).
The "odd readouts" of the Plug-In Tester's Indicator Lamps, when reading these particular Receptacles, is likely due to the damage inflicted by the Rain Water + NaCl being splashed on them.
Notice the Corroded Brass of the Grounded Conductor's Terminal, on the "Lower" portion of the Receptacle.
The Brighter the Green Corrosion is, the fresher the Corrosion, which reduces the overall Conductive areas + increases Contact Resistance.
This could very easily cause the Indicator Lamps (Neon Lamps) to behave oddly.
If Salt concentrations build up within the Receptacle Device, inserting the DVM's Test Leads may have dislodged a chunk of NaCl, which fell across the tips of the Test Leads - resulting in an Arcing Fault.
Since this Fault is not a "Solid" Connection, the Fault Current Level would not fall within the OCPD's Trip Characteristics, and most likely drew no more than 40 Amps for <0.250 Seconds.
This is why you saw lots of Sparks and Arcs, yet there was no tripping of the OCPD.
If the OCPD for the Branch Circuit and the Panelboard's Main have the same Time-Trip Current Characteristics, a Solid Ground Fault will almost always (99.99999555% of the time) trip only the Main Circuit Breaker for the Panelboard, leaving the individual Branch Circuit OCPD closed.
If they both trip, it is because the same Load Current levels are flowing through both Devices at the same time.
Since a Panelboard's Main Breaker will typically have Currents from other Branch Circuits flowing through it, when a Fault occurs on a Branch Circuit, the Main OCPD has the highest Current Level flowing through it, and therefore falls into the Trip Range the fastest.
< end of TCC 101 >
Good luck with this Project.
--Scott (EE)
