a) I don't see an advantage of the conversion other than that the scales may be more comparable (e.g., in terms of their means and standard deviations). But a 10-point response scale is often too much for participants to handle (too many options). It is questionable whether individuals can meaningfully rate constructs in such a fine-grained way. Often, having a 4-, 5-, or 6-point scale is sufficient.

b) There is not much of a difference. The items would still technically be considered ordinal (ordered categorical), regardless of whether they use 4, 5, or 7 response categories, and one should apply methods designed for ordinal variables (e.g., item response models or ordinal factor analysis). With more than 5 response categories, items are often treated as if they were continuous/metrical variables, and that may be fine as long as the items are not too heavily skewed.

I don't agree with Dr Geiser on one point but I consider it important. Ordinal is always ordinal and should be treated that way. David Booth

Scales are usually created from sets of items, so you will be dealing with summed (or weighted) scores. At that point, it really doesn't matter much how you items were originally scaled, but as you note, you can always look at standardized score if you want a more direct comparison.