3.1r1: 8.10.2, multi-drop support

[jadamcrain]

3.1r1's 8.10.2 over the serial transport from #54, for a device on a drop with other devices. Their addresses are set in serial.otherDevices, at least two.

Steps

• Class 1, 2 and 3 reads go to each other device and to the broadcast address. No frame from the DUT may follow within the response timeout; frames from the other devices are noted.
• A read to the DUT must be answered. Before it, the harness waits for the line to go quiet, so a late reply from another device can't mix with the DUT's response.
• That read is repeated for each RTS delay setting to test (serial.rtsDelays, before/after pairs in ms), and the usual delays are restored at the end, even after a failure. The step says which settings to list: for each delay, every selection where it has four or fewer, otherwise its minimum, maximum and two values in between.

Plugin. Whether the DUT raised carrier or asserted RTS, and when, is checked on the line by the plugin, straight away, while the line's recording is of that exchange:

• verifyNoCarrier
• verifyCarrierTiming
• setRtsDelays, which changes the delays or restores the usual ones

Their defaults throw, and DefaultIntegrationPlugin asks the tester, e.g. to read an oscilloscope.

Configuration. The test runs where serial.enabled and serial.multiDrop are true. The configuration check requires:

• at least two other devices, at ordinary addresses (at most 0xFFEF), neither the DUT's nor the master's;
• RTS settings that parse, with none listed twice.

2.7's 8.10.2 stays not implemented, and its -describe output is unchanged.