| | Re: Consonant clusters nt, nd, rt and rd. (part I)
The alveolar plosives (or stops) /t/and /d/ are very unstable in all varieties of English. What changes from variety to variety is how exactly they are reduced. For example, in a word like "bottle" an allophonic variant may be used - for example a glottal stop [ʔ] in London English, or an alveolar flap in many varieties of US English. Or they may be elided completely, as in many of your examples.
If you Google this you'll find three rules for the elision which pop up all the time:
1) The must be at the end of the syllable (the coda) and not at the beginning. So eg in the sentence "I don't want a cup of tea" I could elide the first two /t/ sounds but not the third : /aɪ dəʊn wɑ:n ə kʌp ə ti:/.
The only exception to this that I can think of is "to" in and expression like "want to" - also pronounced /wɑ:nə/. But the explanation is probably that this is a case of gemination + elision (gemination = when two identical phonemes occur consecutively, they are pronounced as one, slightly longer than usual sound) and the geminated consonant which is elided is considered as syllable final. It wouldn't happen to "to" in other contexts where there was a different consonant preceding - eg "back to work".
2) The /d/ or /t/ must be preceded by a consonant with the same voicing - eg /d/ is voiced, so it can be elided in "handsome /hænsəm/ where it is preceded by another voiced consonant /n/ but not in "bad idea" where it is preceded by a vowel. This also explains you query about the -ing forms. the - ing is irrelevant. Look at your examples : they all have /n/ (voiced) before the (unvoiced) /t/ in the word. Elision would be possible I think in a word like "posting" or "wasting" where unvoiced /s/ precedes /t/.
However, the rule doesn't seem to hold up completely : in the sentence quoted before "I don't want a" the /t/ sounds (which are unvoiced) are also preceded by /n/ - yet are elided. In fact, this happens in all negative contractions. and various people have pointed out that /d/ is unlikely to be elided before /r/ (also voiced) , as in "fried rice".
3) It can't happen before /h/. I'm dubious about that that one - I can imagine saying "not him" and eliding the /t/. I admit a glottal stop would be (in my accent) more likely though.
So I'm not fully convinced by any of these rules. I'd therefore go with the account in Gimson "An Introduction to the pronunciation of English". He discusses these rules, but not as absolutes, saying :
"Such elision appears to take place most readily, in rapid speech, in the sequence continuant* consonant + /t/ or /d/ ... followed by a word** with an initial consonant - eg next day, raced back" /neks deɪ/ /reɪs bæk/
"Similarly,word final clusters of plosive or affricate (eg /pt/... /ʤd/) may lose the final alveolar stop when the following word has an initial consonant - eg kept quiet, ... urged them" /kep kwɪəjət/ /ɜ:ʤ ðəm/
"Elision of final /t/ or /d/ is rarer before /h/..."
"Final sequences /nt/ /lt/ tend to keep either /t/ or [ʔ]..."
The use here of "tend to", rarer" etc seem to be to be far closer to reality than the absolute rules found elsewhere.
* a continuant consonant is one you can continue saying as one continuous sound till your breath runs out - eg /s/, /n/