There was never a /ŋ/ sound in the Japanese sound system, and originally /n/ only occurred at the beginning of a syllable. In fact Japanese is quite like Italiano in that for native words all syllables must end in vowel. The kana for syllabic n ん・ン appeared much later and is not included in 五十音図 or 伊呂波歌 and that is saying something. Before the contact with Chinese, Japanese had a very simple syllable structure not unlike that of Polynesian languages. The 撥音(ん), 促音(っ) and 拗音(きゃ/きゅ/きょ/etc) all developed under the Chinese influence.
As we know, Middle Chinese syllables can end in m, n, ŋ, p, t, k. So accommodations must be made for Chinese loanwords. For -p/t/k, a weak vowel (i/u) is added. For -m and -n, み/む is used at the early stages (e.g. 文＝ふみ) and later on ん after its invention (文＝ぶん). As for -ŋ, an extra syllable (a/o/u+う, e/i+い) is used to emulate the Chinese sound the best they can. So in a sense 長音 is also a Chinese influence (there are later sound changes in Japanese, i.e. au/ou->ō which happened during classical times, and ei->ē which is ongoing).
The examples in #3 (行 客 白 gyou gyaku byaku > kau kaku haku), as I understand, happened in Chinese rather than in Japanese. Go-on (呉音) is based on the Wu region (around Shanghai-Suzhou-Nanjing) dialect of the Northern and Southern dynasties (南北朝) period. Kan-on (漢音) is most likely based on the official language spoken in the capital city of the Tang dynasty, Changan (present-day Xi'an). So it's really a difference between Early Middle Chinese and Late Middle Chinese (for example, the loss of
voiced obstruents is clearly reflected in the two readings).
By the way, there are phonetic loanwords in Chinese, from the ancient ones like 葡萄(grape), 琵琶(pipa), 佛(Buddha) to the more modern ones like 沙發(sofa), 吉普(jeep), 維他命(vitamin) though the number is far less than it is in Japanese - probably because we haven't been occupied by Americans, but also probably because our ancestors stopped short of developing an alphabet/syllabary.
