OK, I thought this would be an easy to answer question, but it seems there's not a common terminology for this (time to make it????), so I'll try something that more than an answer should be taken as an introduction to an answer.
I think Roger Atrill's answer is very close, but still limited by the boundaries of specific environments, and not common to other people. Hence, the logical path to create such ontologies (thank you Jonny!) has to be explained every time, something common in Agile environments and the likes, but really cumbersome when you have lots of stakeholders in different locations.
So, my take on this is that there's clearly a group of users that can be qualified and even quantified, which is the group related to the production side. From here, it's easy to define roles and goals, just like Roger says. For the purposes of this answer, let's call this type the "builders group"
There could be an additional type added to the above group, which is the stakeholders that are more involved into the business side, including investors, marketers, administrative staff and such. This is clearly a type of its own, let's call it "financial group", which works and interacts with the builders
Then we have users that really have no involvement with the project, but interact with it in deep ways. For example: government, technology providers, tax office, market regulators, etc. Let's call this type the "external control group": they interact, they have influence, yet they're not part of the company, hence they have no defined roles.
Now, let's go with the user by antonomasia: the consumer, the one that pays the bills, the end user if you like . While some elements of this subset could belong to any of the types above at a certain point, we can say most elements of this group are NOT related to the organization. Hence, there's not a defined role (but there'll be goals for sure). In my simple case, this would be
someone that checks the website that the finance group financed, the
builder group built and the external control group allowed.
For the purposes of this answer, let's simplify this type as "end user".
And now we could be happy, right? But.... there are different types of end users as well! Back to this simple example, one end user could be a member of the site, another could be someone that is looking for something specific coming from a SERP, therefore having a dual state until he engages (or not). Another one is just someone coming from hyperlinking, totally uninterested and which we could call a "bouncer". Someone could be doing a trial, someone could be paying, someone could be testing. Someone could be logged into the system, someone could be logged via social media. So more than goals, we're looking at levels of engagement that could include goals or not
And then, we have.... machine users! Spiders, crawlers, AI, scanners, bots and so on. Someone could say "ok, users are humans so machines are not users", which would make no sense every time we put some much effort to please or fight against these machines in order to make a project successful
Please note one thing: they all are present in the simplest example, so it's clear that we all talk about these groups and types in most projects, yet we have to define their names by their roles and goals, by their organizational title, by personas... And I would also add by their level of engagement or whether they're humans or machines.
In short, it seems there's not a general consensus for something that is at the core of any project (I did search for this answer and asked many UX professionals before asking here), which is particularly strange, specially on a discipline like UX that loves to create names for everything. I propose something along the lines of Influence Types, which could be measured in a relatively easy way, where goals, roles, engagement levels, etc are variables that will define the user more accurately (more or less a process like when in Agile you define points)