[2.0.x] Configure stepper drivers per axis

[teemuatlut]

Builds upon #11133

• Add axis specific driver definitions to determine proper timing parameters like step pulse width, delay after asserting dir level and maximum stepper frequency.
• The user can use settings in Configuration_adv.h to override the parsed driver IC defaults.
• The new driver definitions are used to determine which TMC or L6470 are in use and need initialization.
• By default the Allegro A4988 drivers are enabled for X, Y, Z and E0. More on this below.
• Add new TMC macros that work by checking for a certain feature, rather than the dev checking for a list of driver types that support the feature.
• References to TRAMS removed and will simply be handled by defining the four axis as TMC5130. However this PR doesn't yet move Marlin to the library that supports the driver.
• Differentiating between TMC2130 in SPI configured mode and standalone mode is done by having two driver definitions; TMC2130 and TMC2130_STANDALONE. The way I see it, the default use case for that driver is in configured mode and using standalone mode is a special case. In standalone mode only the stepper timings are affected and they pretty much act the same as TMC2100 as far as Marlin is concerned.
• Updates Travis to use the new driver definition method so that they actually are tested.

---

I decided to use the Allegro drivers as the default to match what is the most common configuration in the user base. Specifying any drivers is not forced but will default to DRV8825 configuration, except for DIR delay. This could be improved, or force defining at least one driver (unless overriden in _adv.h).
I think the main question is if we know the parameters different drivers require, which should be used as the default? I think we should not use 0 delays as there is no driver compatible with such timings.
However, this doesn't take into account the natural delay there is no matter what we do.
This also leads to a more general configuration change that affects the stepper ISR. Perhaps we should no longer check if MINIMUM_STEPPER_PULSE is defined, because from hereon it always is, and rather check if the defined pulse width is greater than 20 for example. @ejtagle ?

---

The compile time driver IC macros:

AXIS_DRIVER_TYPE(A, T)
Check if an axis A has a driver type T associated with it.

HAS_DRIVER(T)
Check if the user has driver type T in any of the axis enabled.

HAS_TRINAMIC
Check if any supported Trinamic drivers are enabled in any axis.

AXIS_IS_TMC(A)
Check if the specified axis A has a supported Trinamic driver enabled.

[GMagician]

@teemuatlut may I close my pull request?

[teemuatlut]

Yes, I believe so.

[ejtagle]

@teemuatlut : This PR does not impact the stepper ISR, i have no problems with it.
I still have to do fine tuning of timing on the stepper, but that will not affect this PR, so no problems

[teemuatlut]

Currently in upstream we have a condition for the step width: #if MINIMUM_STEPPER_PULSE.
I think this condition can be removed as from now on we will always have it defined.
My reasoning then was that we will then also always have a defined delay compiled in, contrary to how it currently is.

[thinkyhead]

Is this sanity-check obsolete already, or is it just now becoming obsolete with this PR?

[teemuatlut]

It's mostly just shuffling the code around to bring more sense into it. Nothing is being removed.

[thinkyhead]

Does this TMC_Z_CALIBRATION test get replaced elsewhere, or is TMC_Z_CALIBRATION being dropped?

[teemuatlut]

Still there.

[thinkyhead]

I see in a few configurations that ENDSTOP_INTERRUPTS_FEATURE is being disabled. Is this because the pins that go with these configs are not interrupt-capable?

[teemuatlut]

Not intentional. I'll need to redo the search&replace operation.

[thinkyhead]

Overall it looks very good. It's much better than the previous method of configuration, with the details now hidden from the user and a much more expressive connection between stepper drivers. This will get merged pretty soon, methinks.

[tcm0116]

@teemuatlut - FYI it looks like the new DRIVER_TYPE definitions weren't added to the generic delta example configuration as part of this PR.

[teemuatlut]

So it seems. Must be because for whatever reason the endstop inverting is turned on by default. I don't know why that should be the case.
I do all config file updates with regex replacements and that's why the file wasn't picked up.

[thinkyhead]

509e1cf