Update concurrency.md

Minor grammar fix with the word " a".

Author: JonathanMcCormickJr

src/part-guide/concurrency.md

@@ -160,7 +160,7 @@ Two computations happen in parallel if they are literally happening at the same
 
 To restate, two computations may happen one after the other (neither concurrent nor parallel), their execution may be interleaved on a single CPU core (concurrent, but not parallel), or they may be executed at the same time on two cores (concurrent and parallel)[^p-not-c].
 
-Another useful framing[^turon] is that concurrency is way of organizing code and parallelism is a resource. This is a powerful statement! That concurrency is about organising code rather than executing code is important because from the perspective of the processor, concurrency without parallelism simply doesn't exist. It's particularly relevant for async concurrency because that is implemented entirely in user-side code - not only is it 'just' about organizing code, but you can easily prove that to yourself by just reading the source code. That parallelism is a resource is also useful because it reminds us that for parallelism and performance, only the number of processor cores is important, not how the code is organized with respect to concurrency (e.g., how many threads there are).
+Another useful framing[^turon] is that concurrency is a way of organizing code and parallelism is a resource. This is a powerful statement! That concurrency is about organising code rather than executing code is important because from the perspective of the processor, concurrency without parallelism simply doesn't exist. It's particularly relevant for async concurrency because that is implemented entirely in user-side code - not only is it 'just' about organizing code, but you can easily prove that to yourself by just reading the source code. That parallelism is a resource is also useful because it reminds us that for parallelism and performance, only the number of processor cores is important, not how the code is organized with respect to concurrency (e.g., how many threads there are).
 
 Both threaded and async systems can offer both concurrency and parallelism. In both cases, concurrency is controlled by code (spawning threads or tasks) and parallelism is controlled by the scheduler, which is part of the OS for threads (configured by the OS's API), and part of the runtime library for async (configured by choice of runtime, how the runtime is implemented, and options that the runtime provides to client code). There is however, a practical difference due to convention and common defaults. In threaded systems, each concurrent thread is executed in parallel using as much parallelism as possible. In async systems, there is no strong default: a system may run all tasks in a single thread, it may assign multiple tasks to a single thread and lock that thread to a core (so groups of tasks execute in parallel, but within a group each task executes concurrently, but never in parallel with other tasks within the group), or tasks may be run in parallel with or without limits. For the first part of this guide, we will use the Tokio runtime which primarily supports the last model. I.e., the behavior regarding parallelism is similar to concurrency with threads. Furthermore, we'll see features in async Rust which explicitly support concurrency but not parallelism, independent of the runtime.