Added docs for state services

packages/sequencer/src/state/StateServices.md

@@ -0,0 +1,32 @@
+## State Service Architecture
+
+The state services are built using a abstraction stack, where every service that is below the root is based on the parent.
+All of this services have the ability to base off one another to allow for diff-based changes.
+You can think of it a bit like git, where every commit is only the diff between the parents state and the next state.
+
+*But*, in this architecture there is no such thing as branching, it is strictly linear and behaves almost like a traditional stack.
+That means, for example, that you shouldn't change state in a service that has one or more children, as these changes might not get reflected instantly in the children's perception of state.
+
+Now I will go over all the different kinds of stateservices
+
+Btw all of this also applies to to merkle tree stores
+
+### AsyncStateService
+
+This is always the base for the whole stack and is meant to be implemented by the actual persistence layer. 
+That means primarily the DB (but is also implemented by the in-memory state service).
+It's functions are async-based in order to enable external DB APIs
+
+### CachedStateService
+
+It receives a AsyncStateService as a parent and can build on top of it.
+It "caches" in the sense that it can preload state entries from the parent (asynchronously) and then lets circuits among others perform synchronous operations on them.
+It additionally caches write operations that can then later be merged back into the parent (or thrown away). 
+
+### SyncCachedStateService
+
+These are the same as CachedStateService, with the difference that it accepts a CachedStateService and requires no preloading.
+
+### PreFilledStateService
+
+Pre-filled with only a part of the data needed. This is mostly used in Task implementation where all state needed is passed as args.