1) Feature 1:
Most users are entering a thread and read it by order.
Due to that, this feature will be enough for most users.
The implementing cost is very cheap.Due to this combination this can be prioritized high.
2) Feature 2:
As I've already wrote... This is a more comlex and covers more options.
As it is more complex its implementation cost is higher. In this case, I guess that it is not simple.
Who will benefit from it?
As I presume that all of the features discussed here will be limited for the non-supporters only for two folders (Bugs and Suggestion), and that the supporters are probably a small group.
Due to that we can say that for supporters this feature will be a nice improvement (at least for some of them), but for the non-supporters it will almost won't change anything.
Due to that the total benefit is not that big for the users as a whole.
Summarizing the two issues - cost and benefit - we can conclude that this will very posibly won't be with a high priority, and in addition, it is a very reasonable sequel feature for Feature-1 that will be implemented before.
The developers job is not always choosing the best option, but choosing those that will give the most benefit per working time.
3) Feature 3:
I believe that it will not cost much above the effort that is needed for Feature-2.
On the other hand, the gain compared to that feature is questionable.
In case it will just bundle all marked-as-unread messages, it will be less convenient to look for messages related to a specific thread.
In case it will give ability to sort them upon thread (topic) and/or others, the "price" rises.
4) Summarizing;
When thinking on the options for the suggested features, they will think not only on the benefit of each feature but also on the following issues:
a) How much will it improve the game / Forum / whatever...?
b) Who will benefit from the feature?
c) What is the cost for the feature in working time?
d) In case implementing a feature can it be done in two phases?
In case all is irrelevant in this case, no one argues that Feature-3 with the ability to sort is the best option.
But, as they are probably a main issue in their consideration, I guess that they will probably will start with Feature-1, continue later on to Feature-2, and when time will allow they will implement the best option defined above.