I haven't personally observed this phenomenon with clicks, but there are well-known cases of consonant epenthesis that are explained in articulatory terms. For example, it is common to observe an epenthetic [t] between [n] and [s], as in the word chance, since the early closure of the nasal passage turns the nasal stop into an oral one before the air gets released for the fricative.
To address the second part of the question, this isn't the sort of situation in which we'd expect contrastive clicks to arise.
It is common to name phonemes according to the form in which they most often surface, out of convenience and convention. So, if phoneme X most often surfaces as [ɭ] but optionally as [ɭ!] in a certain environment--say before non-high vowels, we name that phoneme /ɭ/. We might formulate a rule that says /ɭ/ is optionally realized as [ɭ!] before non-high vowels (or optionally insert [!] between /ɭ/ and non-high vowels). If the next-generation language learner analyzed the [ɭ!] as being the required realization before non-high vowels, we might consider [ɭ!] to be the more common realization of phoneme X and therefore call the phoneme /ɭ!/ and reformulate our rule to say /ɭ!/ is realized as [ɭ] before high vowels (and as [ɭ!] elsewhere). Over successive generations, the [ɭ] part of [ɭ!] might even be lost, so the rule would become /!/ is realized as [ɭ] before high vowels (and as [!] elsewhere). In this situation, we could reasonably say that [!] contrasts with "null" (i.e. the lack of [!]) in as much as the hypothetical words [!ʌm] and [ʌm] would mean two different things. And it would contrast with other (non-ɭ) consonants.
But notice that in all of these different cases, the phoneme inventory itself hasn't really changed. Just the allophones of phoneme X and what we choose to call its underlying form have changed, since [ɭ] and [!] or [ɭ] and [ɭ!] would not be contrastive.
As you probably know, for [ɭ] and [ɭ!] to be considered respective realizations of contrastive phonemes, we'd need to see evidence of minimal pairs--for example a lexicon in which [kʊ˞ɭʌm] means one thing and [kʊ˞ɭ!ʌm] means something else. This situation is unlikely to arise spontaneously under the circumstances you're describing.