I dont understand the problematic part. Why players don't seem to have agency in this?
Firstly, I didn't say that there was a problematic part. I gave recommendations for helping aligning goals. The difference is that I wasn't making a value judgement, but acknowledging that people have different values and how you might use a milestone system in a slightly different way to help align if it was a problem for you.
Secondly, agency is reduced when you take choices away from players, but also, and this is important, agency is not a universal good. Games by their nature restrict agency in myriad ways. How and why agency is restricted is often much more important than the amount of it. Here, agency is reduced because what gives rewards shifts from an openly known and fixed set of circumstances that the players can then choose to pursue to only those things that the GM selects. This isn't, again, bad on it's face, but it is slightly less agency -- the GM has intruded into the decision space and wields fiat.
Look, in your example, there was nothing that prevented players from following character goals over prioritizing earning XP from combats. The decision space was free, the rewards were known, and players could choose. They did choose, but it lead to a game that wasn't fulfilling for you and presumably at least some of them. So you made a change, and instead of combat being the path to the reward, you select what path is available. The players still have the same options -- they can still murderhobo, but they will no longer earn rewards for it. So, in that sense, their agency isn't altered. However, in the prior mode of XP for combat, the triggers for earning XP were clear and well known and so the players had full knowledge of what they could do to earn XP. When it moved to the GM's side (and milestones are usually obscured from the players so as to give the GM better control over pacing by controlling the timing of leveling) they lost knowledge of exactly what was needed, and instead knew that if they followed the hooks they'd get the levels. This left less direct incentivization to play -- they will be afforded opportunities by the GM to achieve the GM's goals and gain levels, so there's less need to push towards getting the rewards. They no longer need to push as hard. This opens up space to pursue less tangible play rewards because the rat race is over and you can indulge in play that pleases aesthetically rather than drive towards the less pleasing XP triggers for combat (less pleasing being assumed for your group based on your description -- other groups can find combat the most pleasing and would react differently). This softening and loosening means that decisions are less important to outcomes, in general, so agency is reduced. Again, I don't think this is a bad thing at all, agency should not be prioritized without reason and intent.
So, I think it's probably a really good choice on your part to move to milestone leveling for your group. It seems like this better aligns your game to what your table enjoys. Agency isn't a fixed good -- it should be priced according to what it can achieve. Here, I think that your choice to slightly reduce player agency by adding in more GM agency to direct the game works to create a better game for you. That should always be the goal, not some myth perfect pursuit of agency.
Agency is a useful thing to look at in game design, because it does have input into value judgements by players. Some players value it more than others. Further, we can look at a game and do a pretty solid analysis of how agency operates in that game by looking at where choices that do things operate and what they can do. However, that analysis cannot tell you if this is a good game for you or not -- only you can tell that. I like some games with lots of agency quite a lot, and others with lots of agency not so much. I also enjoy some games with little agency. Agency should not be the touchstone it appears to be -- it is not a universal good thing to have more agency.