0) You can just soak the material in a solvent, but I personally got the best results basing the material with lime dry tek syle (you could use carb too) and pulling into solvent for a simple total alkaloid extract. I use 91% IPA because it's cheap and readily available.
1) Start with the families we know contain useful alkaloids, the grasses and the legumes are prime candidates. Though in the end the more variety we can check out the better, test everything you can.
2) You technically only need a few mg of plant material to run a sample. However I usually like to use at least 5-10 grams and do a simple drytek extraction (see #0).
3) Reagents are used for follow up. When you get a spot that appears to match the rf for your chosen alkaloid you can verify whether or not it is by dropping the reagent on it and seeing if it gives a positive reaction.
4) Any part really, though for a simple yes/no for alkaloid content it's probably best to sample from a homogenized mixture of the whole plant if possible. Then after confirmation you can narrow down which specific parts may be of interest.
5) Yes that's correct, once you have a spot at an rf that appears to match your alkaloid of interest, the reagent is used for verification.
Also, re: your extraction. Did you only use naphtha or similar selective solvent to make your pulls? If so, perhaps a less selective solvent would have given better results.