Is it good to have all the PhD publications in one journal

For the PhD itself it probably doesn't matter a lot, but if you are looking for a further research career, I'd advise diversifying a bit. It has the following advantages:

• Other people may have another "favorite" journal, and if you have published in multiple journals, you increase the chances of people thinking "wow, she published in xyz!".
• You will get a broader readership and generally more visibility if you publish in different journals.
• There is some risk that people may see your research as too narrow if you only publish it in one single journal.

Maybe you can try adjusting the framing of your research problem a bit, so that another journal will become a better fit for it, and then try to convince your supervisor by comparing the scopes of journals and arguing why this paper fits better to the other journal.

Welcome to academia@stackexchange, @QuantumGirl. It might differ (slightly) between fields, but in general, I'd say it is not good. Journals have different readerships, and you want to reach different audiences. Next to that, you want to avoid the impression that you are a one-trick-pony with research that is so narrow that it is only publishable in one specific journal (again, not knowing the specific circumstances in your field of course). But in general, a bit of diversification to me sounds a much better strategy, even if it is a good journal.

I don't think it will affect the perceived quality of your work if the reputation of the journal declines. However unlikely to occur, it also wouldn't matter, Impact Factors are usually calculated based on the last 3-5 years and can be tracked over time. IF is a crude measure of quality and not many researchers judge an article or journal's quality solely on them. If it's a well-known journal in your field, the research community won't forget it's reputation (at the time you submitted) overnight. By that time your work should stand on it's own anyway, having lead to citations, grants, or further positions in the time being.

I'd be more concerned with the audience you want your work to reach and what you're co-authors can agree on. As my research is interdisciplinary (Bioinformatics), I'd aim at a range of journals for works more related to each field, reaching an audience that may appreciate it, cite in in future works, or use my supporting data and software packages. This has the added benefit of showing my range of skills to an employer just as I would support them with conference participation, teaching experience, or acquiring grants.

This may not be the best approach for everyone, it depends what you are doing and wish to in the future. Generally, you want to publish in field(s) where you wish to work in the future, where-ever that may be. The plus-side is that journals tend to encourage loyalty and will view your next paper as more suitable to their journal if it follows on from or cites existing work in their journal. This would reduce the workload to write for a new audience or defend more thorough review so you can focus on your thesis. However, I'd be cautious of doing this long-term at the expense of publishing opportunities in other journals. It's not common practice to exclusively submit to one journal for the bulk of your career.