A motivation for writing course notes, when textbooks/monographs already exist, is to omit some things, first. This might seem strange, until one observes that many standard texts which go through several editions often succumb to the natural pressure, amplified by publishers, to become ever-more encyclopedic. The latter has its uses as reference, but perhaps not so much as a guide to the actual threads of a live course. Further, _making_choices_ of what to pay attention to in the course, as opposed to what to ignore (in the course), imparts considerable information to the students.
As positive side-effect of omitting some iconic-but-not-immediately-essential things, sometimes there is room to include newer or novel things that otherwise wouldn't find a spot, or happen not to be included in the standard texts.
Of course, one's own notes can more accurately support and/or complement one's own lectures and plot-line for a course, so one could argue that picking a standard text and just indicating a subset of it for students could suffice. Even if that is the case, people will see all the other stuff and wonder why you're ignoring it. Intentional, or accidental? Some distractions.
And self-generated notes can be made freely available, and available in electronic formats, and...
As to "important considerations", ... certainly facility with an adequate typesetting software system comes first, or it's hopeless. For a first pass through, don't keep everything in one big file, both because it takes longer to do the inevitable frequent re-typesettings, and because it's hard to find things. Use your computer's file system, etc., to organize things. Keep backups. Some form of "version control" can be useful.
About content: especially if you're trying to lighten the burden of encyclopedic texts, don't fall into the (sometimes appealing) trap of "completeness", or of giving all the iconic examples, or perhaps even working out all details of all proofs of all small things. Instead of contriving artificial exercises, if you've given representative prototypes of relevant arguments, subsequent similar arguments can be left to the reader. Of course, one should be honest with the reader, rather than merely lazy, and be honest with oneself about why something's omitted. (Not to mention competent to confirm that it really is just "more of the same" and routine.)
A different sort of point: write at a level and in a style to address the students you actually have, rather than ones you wished you had, or some idealized students. E.g., don't omit arguments or explanations that you know perfectly well most of your students couldn't fill in, don't couch things in terms you know are more sophisticated than your audience, ...
Coming to some questions of taste: the pressures for traditionally published texts to be fairly formal are not entirely helpful to students or mathematics. That is, there is stylistic pressure to be formal, and to follow a certain (quite recent, modern!) style of definition-lemma-proof-lemma-...-theorem-proof... exercises. Candidly, one often discovers that the exercises are cryptic exhibition of the formative examples that motivated the whole business, but the natural causal sequence has been inverted.
Now, yes, a certain amount of that inversion can be helpful, clarifying, but the standard full-blown version is not. (To name-drop) Gelfand was a consistent advocate of giving the key examples so that "definitions and theorems and proofs" become almost afterthoughts. E.g., definitions should be "justified" by examples...
Especially in fairly sophisticated mathematics, terminology is heavy, usage is inconsistent or slightly abused, etc., and trying to "repair" the situation by cranking up the formality is attractive, but often doomed or at least incredibly expensive. Instead, giving the watershed examples, rather than trying to make "rules" or rigidify language, seems to me to more directly address the genuine issues.
And be prepared to defend (on scientific and pedagogical grounds) any deviation from the iconic sources, because it seems that people are quite happy to believe in the authority of iconic authors' writings. Many people in the math business do seem to believe that definitions given in iconic texts are "sacred", too, and are beyond discussion, so don't be surprised when people are inflexible in this regard! Have explanations ready... Yes, some of your energy will be spent defending your choices to students, and this can be made mostly productive, if you are prepared.