If you only have access to one phone number and not giving it out is critical, then Signal might not be the right choice for you. But you won’t find a more secure channel that collects less metadata anywhere else.
There's no property of the latter that makes them a better choice than the former, but the existing ecosystem makes a disposable or role email address a much easier thing to obtain, and in general leaking an email address a far-less-damaging privacy violation than leaking a phone number, which can so easily be used to harass and directly track you.
In many parts of the world, disposable mobile numbers are very definitely not a practical thing. The correct starting point here is to use either an email or a telephone number as an external ID. It's still not perfect, but at least it's not a complete disaster any more.
Alas they've been 'working on it' for a very long time now, and are likely to fail because of this painfully slow progress.
Perhaps if they'd pissed around less with trivia like cryptographically secure stickers, they might have increased their chances of becoming a useful product before they end up surrendering the space to an inferior product which gains too much market share to overcome before they're even properly off the starting blocks.
I criticise Signal here because I like the design and hate their botched execution, not because I dislike the protocol. On the contrary, I dearly want something that competent to succeed, but fear we're rapidly losing the chance of that happening because they have launched a privacy-disaster product and most of the potential market will have seen that, dismissed it and forgotten about it before they pull their finger out and fix it.
There is another unique identifier that's stored in the local contact list: email addresses.
Use either email address or a phone number as an identifier, and you've no longer built a offensively privacy-violating service but have exactly the same distributed property.