variant.cpp has so many mistakes that it can only be intentional

Hi,
I've been using this [variant.cpp](ArduinoCore-samd/variants/arduino_zero/variant.cpp) and the g_APinDescription has so many mistakes in it that I can only conclude these mistakes are maybe intentionnal. Let's take a [random](https://github.com/arduino/ArduinoCore-samd/blob/master/variants/arduino_zero/variant.cpp#L134) example:
```
{ PORTA, 6, PIO_TIMER, (PIN_ATTR_DIGITAL|PIN_ATTR_PWM|PIN_ATTR_TIMER), No_ADC_Channel, PWM1_CH0, TCC1_CH0, EXTERNAL_INT_6 }, // TCC1/WO[0]
```
If you read the spec sheet, PA06 can be used for many things, as correctly noted in the comments
```
| 8 | ~8 | PA06 | | EIC/EXTINT[6] ADC/AIN[6] AC/AIN[2] PTC/Y[4] SERCOM0/PAD[2]
TCC1/WO[0]
```
This pin can be muxed to be
* external interrup 6
* adc in 6
* ac in 2
* ptc 4
* sercom0 2
* tcc1 0

but the description tells e.g. No_ADC_Channel (and omit attribute PIN_ATTR_ANALOG)

Before I spend time fixing this and I submit a PR, can somebody confirm it is not a "feature".
And if it is indeed a feature, I propose to add, starting from e.g. position 50 in g_APinDescription (i.e. leaving a few empty spots), the complete pin definitions for each PortA and PortB pins (even for pins not broken out)

Start with variants/arduino_zero/variant.cpp and inspect g_APinDescription around the PA06 entry, comparing its attributes with the SAMD21 specification table cited in the issue. Confirm whether the existing mappings are intentional; done means documenting that decision or preparing complete, accurate PortA and PortB pin definitions as proposed.

Written by the indexing model from the issue text.