A sound in a language can "actually" be two things. One is, what it actually is phonologically. The other is, what it is phonetically, in this case, exactly now is it articulated? Phonologically in English, it is a single segment. In Czech, it could be a cluster, or it could be a postalveolar affricate (single segment), I don't know of any compelling arguments one way or the other. In the APA that affricate is transcribed as [ǰ], and in the IPA it is (usually) transcribed as [dʒ]. The APA transcription recognizes the unitary nature of the affricate, but the IPA does not recognize the concept "affricate", so you are forced to use a two-symbol sequence. The letter <d> in IPA is ambiguous, standing for any of dental, alveolar and postalveolar. If it is necessary to mark a "t" as dental (as is the case in some Dravidian and Australian languages because there is a dental / alveolar contrast) you can us a diacritic as in [t̪]. There is no symbol or diacritic for a postalveolar stop. The writing <dʒ> does not say that the closure portion is different from the release portion, and in fact there is no unambiguous way to say – using symbol choice – that the two phases of the consonant are at the same place of articulation. If you want to indicate "a unit", you can write <d͡ʒ>, which can be necessary in some languages (when clusters and affricates contrast).
When you transcribe words of a language in IPA, you might only be transcribing phonemically, in which case you only use the minimal symbols for phonemes of the language. In Polish, you have to transcribe trzy "3" differently from czy, where the former has a cluster and the latter has a single consonant affricate, which is where the tie diacritic is pressed into service. As far as I know, Czech does not face that problem. Alternative transcriptions like [ďž] simply are not IPA, so a better question would be, if you're not using IPA, why don't you follow the APA standard and write [ǰ]. The IPA-consistent transcription [ɟʒ] is rejected for two reasons. First, there is no phoneme [ɟ] in Czech. Second, [dʒ] is auditorily closer to the phonetic output than [ɟʒ] is, and the general rule is to pick the closest symbol, consistent with the extant phoneme symbols. The alternative [ɟ] is rejected for "dʒ" because that stands for a completely separate phoneme (the one spelled <ď>).
Sticking only to symbols with phonemic status in Czech, the cluster transcriptions [dɟ, dʒ] as phonetic descriptions both suffer from the same problem of not explicitly saying that the closure and release are at the same place of articulation, you have to add a footnote saying that. [dʒ] is better than [dɟ] given the standard of auditory closeness.