Smoothing and compression in CEPD

On looking up the word diamond in the new Cambridge English Pronouncing Dictionary (CEPD; 18^th edition, 2011), one finds that it is transcribed ˈdaɪə.mənd in British English (p.137). The transcription provided is explained as to be interpreted as conveying that the variants ˈdaəmənd, ˈdaːmənd, and ˈdɑːmənd, which I used in my post on the diamond jubilee last week, are all possible. What this transcription also shows is that, according to the editors, the word in question is usually perceived as being disyllabic by British native speakers. 

If one looks up other varisyllabic words in the dictionary, words that is in which smoothing and compression are both possible (and indeed extremely common) in current RP, one finds a great deal of inconsistency: 

hour is (usually smoothed and monosyllabic) aʊə^r, (less commonly disyllabic and not smoothed) ˈaʊ.ə^r (p.239);

our is (usually smoothed and monosyllabic) aʊə^r, (less commonly disyllabic and not smoothed) ˈaʊ.ə^r, (monosyllabic) ɑː^r (p.355);

nowadays is (usually trisyllabic) ˈnaʊ.ə.deɪz, (less usually disyllabic) ˈnaʊə- (p.342);

shire is (monosyllabic, often heard as smoothed) ʃaɪə^r, (less commonly disyllabic and not smoothed) ˈʃaɪ.ə^r (p.447);

society is (only heard with four syllables and not smoothed) səˈsaɪ.ə.ti (p.457).

Why all this inconsistency? The answer is to be found in the introduction to the CEPD, pp.vii-viii:

“We need to consider the special case of diphthongs followed by a schwa (…). Such sequences are sometimes referred to as triphthongs; they are not phonemes of English, but combinations of diphthongs with the schwa (/ə/) vowel. However, they present unique problems: in BBC pronunciation [aka RP] many of these triphthongs are pronounced with such a small movement in vowel quality that it is difficult for foreign learners to recognize them; for example, the name ‘Ireland’, which is traditionally transcribed /ˈaɪə.lənd/, frequently has an initial syllable which sounds virtually indistinguishable from /ɑː/, with just a small movement towards /ɪ/ and then towards /ə/ at the end. Some triphthongs are pronounced like single syllables, as in the example just given, while others are more likely to be divided into two syllables (…). We usually find the two-syllable pronunciation (i) when the schwa is a separate morpheme (e.g. '-er' in 'buyer’ /ˈbaɪ.ə^r/), (ii) when the word is thought to be foreign (this includes many biblical names originating from Hebrew, e.g. ‘Messiah’ /məˈsaɪ.ə/), and (iii) when a word is not used very frequently, e.g. ‘cyanosis’ /ˌsaɪ.əˈnəʊ.sɪs/. Where we feel a triphthong may be pronounced either as two syllables or as one, we give one pronunciation with a syllable division and an alternative with a one-syllable pronunciation. In the single-syllable case the middle vowel is printed in italic to indicate that in this reduced pronunciation the middle vowel may be very much reduced or even elided (omitted); for example, ‘briar’ /ˈbraɪ.ə^r, braɪə^r/; where the one-syllable pronunciation seems more likely, that pronunciation is given first: ‘fire’ is /faɪə^r, ˈfaɪ.ə^r/. “ 

What does "seems more likely" in the last sentence mean? 

I wonder why power, for example, would be more usually heard as ˈpaʊ.ə (p. 389) but flower more commonly flaʊə (p.191). Why would science be ˈsaɪ.ənts and (less commonly) saɪən^ts (p.437), but scientist only ˈsaɪ.ən.tɪst (p.437)? A bit confusing for foreign learners of English, I think. 

What is the rationale behind these choices? Can word frequency be a valid reason for transcribing the above words as disyllabic in some cases and monosyllabic and smoothed in others?