feat: Zigbee Specification revamp

[Nerivec]

• Move zcl into zspec folder.
• Add first draft for ZDO specification [R23]. This PR doesn't introduce its usage, it needs extensive testing/tweaking before it can be refactored in.
• Add ZCL-specific test (independent of higher-level herdsman logic).
• Reorganized imports/exports for zspec
  • consts/enums will now be automatically exported whenever new ones are added in their respective places.
  • import {ZSpec, Zcl, Zdo} from 'zigbee-herdsman'
  • Zigbee consts => ZSpec.ZIGBEE_REVISION
  • Zigbee enums => ZSpec.BroadcastAddress.RX_ON_WHEN_IDLE
  • Zigbee util functions => ZSpec.Utils.channelsToUInt32Mask()
  • ZDO consts => Zdo.ZDO_ENDPOINT
  • ZDO enums => Zdo.LeaveRequestFlags.WITHOUT_REJOIN
  • ZDO util functions => Zdo.Utils.getServerMask()
  • ZDO status/error => Zdo.Status.SUCCESS and new Zdo.StatusError(Zdo.Status.INV_REQUESTTYPE)
  • ZCL consts => Zcl.POWER_SOURCES
  • ZCL enums => Zcl.DataType.UINT8
  • ZCL util functions => Zcl.Utils.isClusterName()
  • ZCL status/error => Zcl.Status.SUCCESS and new Zcl.StatusError(Zcl.Status.FAILURE)
  • ZCL frame => Zcl.Frame
  • ZCL header => Zcl.Header
• Add enums: ParameterCondition, DataTypeClass and replace hardcoded strings in Clusters/Foundation.
• Fix typing for ParameterDefinition, and more...
• Extract Xiaomi workaround from CHAR_STR to its own data type MI_STRUCT.
• Add several basic consts that should eventually replace hardcoded stuff throughout the code. Can be swapped when encountered in future PRs. More can be added as needed.
• Added several basic Zigbee types as "more informative" wrappers for Typescript types. Same as above.

TODO:

• ZDO buffalo tests
• Update zigbee2mqtt imports
• Zigbee Specification revamp zigbee-herdsman-converters#7488
• Update zigbee2mqtt-frontend imports
• Remove old tests now covered by new ZCL-specific ones (kept unchanged until merge to ensure no change in logic creeps up)

[Nerivec]

@Koenkk If you can review this commit in great details. It drastically changes the code for clusters/custom clusters lookup. We need to make sure it still matches the way you designed custom clusters to be (i.e. that tests aren't testing this wrong). 😅
If it does, these changes have one heck of a nice side-effect:

x10000 getCluster:
Old: ~280ms
New: ~50ms

x10000 getCluster with 2 custom clusters:
Old: ~7.5s
New: ~57ms

[Koenkk]

Thanks!