When it comes to "review" publications, which author is considered more important: the first or the last?

Generally, the first author is the primary researcher, and the last author in the list is the supervisor or lead. The corresponding author, if different from the above two, is also important.  The order of the remaining authors signifies the importance of their roles in the work.

For example, during a PhD, the primary researcher is the PhD scholar who performs most of the research work and writes papers. The supervisor is normally the last author. Sometimes, the supervisor may feel they have done more work, so they ask the student to put their (supervisor's) name first.

However, this is merely a convention used to infer the authors' roles in a publication from their order in the list. Other conventions may be used to clarify their roles, like explicitly declaring the contribution of each author in the paper.

In my view the first one is considered as the most important because authors has to be ordered based on their size of contribution

The names of the authors of research should be arranged according to the size of the contribution to the research. 

Depends here the review is published - For unsolicited review articles, the first author is generally considered to have done most of the work and writing, whereas for invited reviews, at least in high impact journals it would be the first author who is senior (but not necessarily more important!) 

In many cases, last authors of review articles are added for many secondary purposes. For example, a corresponding author add additional names of scholars as authors (not really author of this article) in his/her article so that publish 2 or 3 articles in the future instead of just one article currently ready for publication. This is an ethical issue that should be considered by databases like Web of Science, Scopus, or COPE for application to peer-reviewed journals, and also universities and academic institutions. But if an author adds last authors based on their actual contribution to the review or even research works, they can do so for drafting, reviewing, editing, providing sources and data, and even conceptualizing.

Author contributions should be explicitly stated in a dedicated section at the end of the publication. If the journal template does not include an author contribution section, you can use the acknowledgments section. In the author list at the beginning you can mark equally contributing authors at any place in the list (equally contributing first, equally contributing last, or anywhere in the middle).

Finally, I would refrain from using the term "important" when it comes to authorship. If a person qualifies as author for a publication, it means their contribution has been "important" by definition. For review papers, the typical assumption of first author - junior, hands-on / last author - senior, supervisor, funding does is not as often true. 
But by using author contribution statements, and preferably in one of the standardised formats (such as the CRediT taxonomy), all doubts are eliminated.